Etude de la connectivité entre les communautés de poissons de

Transcription

Étude&de&la&connec,vité&entre&les&communautés&
de&poissons&de&différents&habitats&du&lagon&de&
Nouvelle9Calédonie&par&microchimie&des&otolithes&
et&de&l’environnement&
Christelle&Paillon&
Thèse&de&Doctorat&
Spécialité&Ecologie&marine&
Ecole&doctorale&du&Pacifique&
Soutenue&le&22&avril&2014&
Jury%%%
Payri&Claude& Présidente&du&jury&
Panfili&Jacques& Rapporteur&
Meekan&Mark& Rapporteur&
Chateau&Olivier& Examinateur&
Coutures&Emmanuel& Examinateur&
Wan,ez&Laurent& Directeur&de&thèse&
Vigliola&Laurent& Co9directeur&de&thèse&
Université de la Nouvelle-Calédonie
Ecole doctorale du Pacifique (ED 469)
Thèse de Doctorat
Spécialité Ecologie marine
Présentée par
Christelle PAILLON
Pour obtenir le grade de
Docteur de l’Université de la Nouvelle-Calédonie
Étude de la connectivité entre les communautés de poissons de
différents habitats du lagon de Nouvelle-Calédonie par
microchimie des otolithes et de l’environnement.
Soutenue le 22 avril 2014 à l’Université de la Nouvelle-Calédonie, devant le jury composé
de :
Pr. PAYRI Claude
Dr. PANFILI Jacques
Dr. MEEKAN Mark
Dr. CHATEAU Olivier
Dr. COUTURES Emmanuel
Dr WANTIEZ Laurent
Dr VIGLIOLA Laurent
IRD, CoRéUs 2
IRD, Ecosym
AIMS, Perth
Aquarium des Lagons
Province Sud
UNC, LIVE
IRD, CoRéUs 2
Présidente du jury
Rapporteur
Rapporteur
Examinateur
Examinateur
Directeur de thèse
Co-directeur de thèse
Laboratoires d’accueil et direction scientifique :
Wantiez Laurent (Maître de Conférences, HDR)
EA4243 Laboratoire Insulaire du Vivant et de L’Environnement
Université de la Nouvelle-Calédonie
BP R4, 98851 Nouméa Cedex
Vigliola Laurent (Chargé de Recherche)
Unité de Recherche 227 CoRéUs
Biocomplexité des Ecosystèmes Coralliens de l’Indo-Pacifique
Institut de Recherche pour le Développement
Centre de Nouméa
101 Promenade R. Laroque,
BP A5, 98848 Nouméa Cedex
Remerciements
Je tiens à remercier toutes les personnes qui ont croisé ma route entre la Nouvelle-Calédonie,
la France et l’Australie et qui ont contribué, de près ou de loin, à la réalisation de ce projet.
Je tiens tout d’abord à remercier mes directeurs de thèse, Laurent Wantiez et Laurent Vigliola,
pour m’avoir permis de réaliser cette thèse et pour m’avoir fait confiance pendant ces quatre
années. Je les remercie aussi de leur soutien qui m’a permis de surmonter mes incertitudes et
de relever le défi.
Je remercie également les laboratoires et équipes de recherche qui m’ont accueillie au cours
de cette thèse. Je remercie Claude Payri, directrice de l’unité CoRéUs, Gilles Fédière et
Georges de Noni, directeurs de l’IRD, pour m’avoir accueillie au sein du centre de Nouméa.
Je remercie le Pr Hamid Amir, directeur du LIVE, Jean-Marc Boyer et Gaël Lagadec,
présidents de l’Université de la Nouvelle-Calédonie, pour avoir hébergé ce travail de thèse.
J’adresse mes remerciements à l’ensemble du jury de thèse pour avoir accepté d’évaluer mon
travail. Merci à Jacques Panfili et Mark Meekan d’avoir accepté d’être les rapporteurs et à Olivier Château et Emmanuel Coutures d’être les examinateurs. Je tiens aussi à remercier les
membres de mon comité de thèse : Pascale Chabanet, Maylis Labonne et Dominique Ponton,
pour leur disponibilité et les remarques constructives qu’ils ont pu apporter au cours de ma
thèse. Je tiens aussi à remercier Michel Kulbicki pour son aide précieuse et son savoir.
Je remercie le programme Zonéco (opération « Microchimie des otolithes ») ainsi que la
Province Sud (Prix d’Encouragement à la Recherche) pour avoir soutenu ce travail.
Je tiens particulièrement à remercier l’équipe technique de CoRéUs et du centre IRD Nouméa
sans qui cette thèse n’aurait pu être possible. La team de terrain, Gérard Mou-Tham, terrible
et efficace chasseur et Miguel Clarque, toujours de bon conseil pour l’organisation des
missions. Je tiens à remercier Joseph Baly pour la précision de tes dissections et ton éternel
sourire même lorsque je te sortais une montagne de poissons à disséquer. Je remercie les
différents pilotes du dock océano, Philippe Naudin, Samuel Tereua et Napoléon Colombani
présents aux cours des missions.
Merci aux équipes du Pôle de Spectrométrie de Brest et du LEMAR qui m’ont accueillie pour la mission laser bretonne : Claire Bassoulet, Maylis Labonne, Fanny Sardenne, Eric Dabas et
Jean Marie Munaron. Merci à Olivier Bruguier et Chantal Douchet du laboratoire
Géosciences Montpellier pour les analyses de l’environnement. Merci à Niels Munskaarg,
David Parry et l’équipe du RIEL de l’Université de Darwin qui m’ont accueillie lors des missions laser australiennes. Je tiens particulièrement à remercier Françoise Foti pour m’avoir chaleureusement accueillie sur le sol Darwinien et m’avoir accompagnée de longues heures dans le froid de la salle ICP afin de me laisser les rênes de l’infernale machine.
Cette thèse n’aurait pu se réaliser sans la présence de la team thésards (et associés) de
CoRéUs et d’ailleurs, équipe de collègues mais aussi amis proches pour beaucoup.
Je tiens particulièrement à remercier Laury qui a supporté les multiples crises existentielles ou
euphoriques et avec qui la symbiose était proche de la perfection (si, si !!). Lyly, je pense que
l’aboutissement de cette thèse n’aurait pas été possible sans toi. Merci d’avoir eu la patience de me remonter le moral en toutes circonstances et merci aussi d’avoir rit à toutes mes blagues, même les plus incertaines (et il y en a eu !).
Je tiens également à remercier mes collègues du bureau 60 des thésards CoRéUs, bureau qui a
vu tant de monde passer. Je remercie beaucoup Haizea et Bastien qui m’ont accueillie, guidée
et ont partagé de grandes discussions écologiques (ou pas !). Je remercie aussi le trio des
thésardes plongeuses : Josina, Marion et Stéph pour leurs good vibes et les pauses théscraquages de fin d’aprèm. Merci aussi à Isabelle et Jean-Baptiste, la première et le petit
dernier du bureau des thésards.
Je tiens particulièrement à remercier Magali pour sa précieuse aide en statistiques, pour
m’avoir convaincue de me mettre à l’apnée et pour les parties de coinche endiablées jusqu’au petit matin.
Je remercie Delphine, Marine, Andres, Christophe M, Simon, William, Christophe C et
Mélanie, Sylvie, Jérôme, Fréderic, Hervé, Houssem, Matthieu, Agathe, Christophe V,
Daphnée, Marilyn, Charles et autres (désolée si j’en oublie !) pour les discussions animées au
détour d’un couloir, autour d’un café ou lors d’une apéro-thérapie.
Je remercie le « Lama et associés » pour la pause détente du matin : Léo, Stéph, Jacob, Anne,
Luc, William, Adeline, Hélène, Anne-So, Denise et Jenny.
Je ne pourrais pas tous les citer (ou re-citer) mais je tiens particulièrement à remercier mes
amis du Caillou qui se reconnaitront. Merci de m’avoir accompagnée et supportée durant cette
aventure, et merci d’avoir accepté mon absence durant la fin de cette thèse. Merci aussi à mes
amis Normands, Montpelliérains, Lillois et Wimereusiens qui ont été présents malgré la
distance.
Pour finir, je tiens particulièrement à remercier ma famille qui m’a soutenue à travers
cyclones et raz de marées et sans qui cette thèse n’aurait jamais pu se faire. Maman, Papa,
Patrick, Julie, Tanguy et Tiphaine, je ne saurais comment vous remercier pour votre
indéfectible soutien.
« Necesito del mar porque me enseña… »
Extrait de El Mar
Poème de Pablo Neruda
Sommaire
Chapitre 1 : Introduction générale ......................................................................................... 1
1. La connectivité............................................................................................................................ 3
2. Méthodes de mesure de la connectivité ...................................................................................... 9
3. Microchimie des otolithes......................................................................................................... 12
4. Contexte en Nouvelle-Calédonie .............................................................................................. 16
5. Objectifs de la thèse .................................................................................................................. 17
Chapitre 2 : Méthodologie générale ..................................................................................... 19
1. Sites d’étude ............................................................................................................................. 21
2. Plan d’échantillonnage.............................................................................................................. 26
3. Techniques de prélèvement ...................................................................................................... 29
4. Protocole de préparation des échantillons ................................................................................ 31
5. Analyses microchimiques ......................................................................................................... 34
6. Traitement des données ............................................................................................................ 39
Chapitre 3 : Quantification du pouvoir discriminant des signatures multi-élémentaires
mono-spécifiques et multi-spécifiques dans le lagon de Nouvelle-Calédonie.................... 45
Manuscrit A: Assessing the power of otoliths multi-elemental signatures in the coral reefs and
mangroves of New Caledonia .......................................................................................................... 49
1. Introduction .............................................................................................................................. 51
2. Material and methods ............................................................................................................... 53
3. Results ...................................................................................................................................... 59
4. Discussion ................................................................................................................................. 66
5. Conclusion ................................................................................................................................ 69
Chapitre 4: Comparaisons entre les signatures multi-élémentaires de l’environnement et des otolithes en Nouvelle-Calédonie...................................................................................... 73
Manuscrit B: Comparison of otoliths and environmental multi-elemental signatures in the coral
reefs and mangroves of New Caledonia .......................................................................................... 76
1. Introduction .............................................................................................................................. 78
2. Material and methods ............................................................................................................... 80
3. Results ...................................................................................................................................... 86
4. Discussion ................................................................................................................................. 93
5. Conclusion .............................................................................................................................. 101
Chapitre 5 : Importance du rôle de nurserie des mangroves de Nouvelle-Calédonie pour
un poisson corallien, Lutjanus fulviflamma ........................................................................ 105
Manuscrit C: Essentialness of mangroves as nursery habitat for a coral reef fish in New
Caledonia. ....................................................................................................................................... 108
1. Introduction ............................................................................................................................ 110
2. Materials and methods ............................................................................................................ 113
3. Results .................................................................................................................................... 119
4. Discussion ............................................................................................................................... 123
Chapitre 6 : Connectivité et patrons de mouvement d’une espèce de poisson corallien,
Lutjanus fulviflamma ............................................................................................................ 131
Manuscrit D: Migratory paths and connectivity for a coral reef fish, Lutjanus fulviflamma. ... 135
1. Introduction ............................................................................................................................ 137
2. Materials and methods ............................................................................................................ 140
3. Results .................................................................................................................................... 147
4. Discussion ............................................................................................................................... 152
5. Conclusion .............................................................................................................................. 156
Chapitre 7 : Discussion générale et perspectives ............................................................... 161
1. Axe méthodologique............................................................................................................... 163
2. Axe écologique ....................................................................................................................... 165
3. Conclusion et perspectives de travail...................................................................................... 167
Références bibliographiques ............................................................................................... 171
Liste des figures .................................................................................................................... 193
Liste des tableaux ................................................................................................................. 197
ANNEXES ............................................................................................................................. 199
Chapitre 1 : Introduction générale
1
2
Les paysages récifaux-lagonaires sont des mosaïques d’habitats discontinus distribués
sous forme de patchs et représentés par les récifs coralliens (barrière récifale, récifs
intermédiaires et frangeants), les herbiers, les fonds meubles, les mangroves, etc. (Jones et al.
2009). Ces paysages sont considérés comme des oasis de diversité et de productivité dans les
déserts océaniques. Ils font partis des écosystèmes les plus diversifiés de la planète, et sont
ainsi souvent comparés aux forêts tropicales humides en terme de diversité spécifique
(Bellwood & Hughes 2001). Ils possèdent une valeur économique inestimable et des
populations entières vivent grâce aux biens et services qu’ils fournissent (e.g. pêche, tourisme,
etc.). Cependant, ils sont considérés comme un des écosystèmes marins les plus impactés car
à l’échelle mondiale, 30% des écosystèmes récifaux sont d’ores et déjà considérés gravement perturbés tandis que 60% seront dramatiquement menacés à l’horizon 2030 (Hughes et al.
2003, Burke et al. 2011). Les origines de ces maux sont diverses, les récifs coralliens sont
affectés par de nombreuses perturbations d’origine naturelle et anthropique qui peuvent être de nature épisodique (e.g. tempêtes, exposition à des polluants toxiques, etc.) ou chronique
(e.g. surpêche, eutrophisation, aménagement de l’environnement côtier, exposition aux
activités minières, etc.) (Nystrom et al. 2000, Pandolfi et al. 2003, Wantiez et al. 2006,
Sandin et al. 2008). Alors que les perturbations naturelles constituent des opportunités de
renouvellement et de développement des récifs coralliens, l’augmentation de leur fréquence ou de leur intensité associée aux changements globaux et l’intensification des perturbations
anthropiques sur ces écosystèmes remet en question leur capacité de résilience.
Parmi les écosystèmes associés aux récifs coralliens, la mangrove est un écosystème clé à
l’interface entre le milieu terrestre et le milieu marin. Elle fait partie des écosystèmes les plus productifs au monde et possède une forte valeur, aussi bien au niveau écologique
qu’économique. La mangrove possède un rôle biologique important comme habitat, zone de
nurserie, d’alimentation ou de refuge pour de nombreuses espèces et notamment des poissons
et invertébrés comestibles (Nagelkerken et al. 2008). Ce rôle biologique est donc à l’origine d’une ressource alimentaire importante pour les populations humaines (Mumby et al. 2004).
De plus, la mangrove possède un rôle physique majeur en protégeant le littoral de l’érosion ainsi que des événements climatiques violents (e.g. tempêtes, raz de marées). Elle agit comme
une zone tampon en piégeant les particules issues des apports terrigènes et permet ainsi le
1
développement des récifs coralliens adjacents dans une eau de faible turbidité. Au cours des
cinquante dernières années, 35% de la surface mondiale des mangroves a été perdue (Valiela
et al. 2001). Cette surface, actuellement comprise entre 160 000 et 200 000 km2, disparait à
un taux moyen d’1 à 2 % par an sous l’effet de multiples menaces anthropiques pesant sur cet écosystème (Spalding et al. 1997, Nations 2007). Le rôle physique de la mangrove est menacé
par l’urbanisation accrue et l’expansion des activités industrielles entrainant une déforestation massive et une conversion des zones de mangrove en terrains agricoles, aquacoles ou en zones
constructibles. De même, l’augmentation des rejets d’effluents d’origine anthropique surpasse leur capacité de tampon et entraine une eutrophisation du milieu. Enfin, l’exploitation à outrance des ressources naturelles menace le rôle écologique de la mangrove.
Associée au régime de perturbations naturelles, la pression anthropique exercée sur les
mangroves et les récifs coralliens entraine de nombreux effets néfastes dont une diminution de
la biodiversité de ces écosystèmes, une modification de la structure des communautés, un
changement dans la structure de l’habitat voire une perte d’habitats (Nystrom et al. 2000,
Valiela et al. 2001, Alongi 2002, Pandolfi et al. 2003, Bellwood et al. 2004).
La structure en patchs des paysages récifaux-lagonaires induit une fragmentation des
écosystèmes et des espèces associées. Ces dernières s’organisent en métapopulations comprenant de nombreuses sous-populations ou populations locales de différentes tailles. La
fragmentation des populations, l’érosion de l’habitat et l’augmentation des perturbations anthropiques que subissent les récifs coralliens et les mangroves en font des écosystèmes
côtiers extrêmement fragiles où les connections entre patchs par dispersion ou migration des
individus sont indispensables à leur fonctionnement (Mumby & Hastings 2008). Déterminer
l’étendue des liens entre les populations locales reste un grand challenge et demeure d’une importance capitale pour l’étude et la préservation des récifs coralliens et des mangroves. En
effet, la connectivité entre populations locales a d’importantes implications dans la persistance et la résilience des populations face aux perturbations naturelles et anthropiques.
De plus, l’estimation de la connectivité entre populations locales est un objectif clé pour la
préservation et la gestion des espèces car le design des réseaux d’aires marines protégées (taille, lieu et espacement) ne peut être optimal sans une estimation préalable de la
connectivité existante (Jones et al. 2007, Almany et al. 2009, Lowe & Allendorf 2010, Mora
& Sale 2011).
2
1. La connectivité
1.1. Définition générale
La connectivité est liée au degré de mouvement des organismes, d’échange de matière ou d’énergie au sein d’un paysage (Crooks & Sanjayan 2006). Il existe deux composantes
principales dans le concept de connectivité : 1) la composante structurelle (physique) qui
réfère à la propriété de la structure du paysage à faciliter/entraver les mouvements des
organismes et 2) la composante fonctionnelle qui correspond à la réponse comportementale
des organismes face à cette structure physique (Fig. 1.1). Cette composante fonctionnelle est
elle-même subdivisée en deux catégories : la connectivité potentielle englobant les
connaissances de base acquises sur les capacités de dispersion des organismes et la
connectivité réelle qui va traduire une quantification directe du mouvement réel des individus
au sein d’un paysage donné (Calabrese & Fagan 2004, Moilanen & Hanski 2006).
Figure 1.1 : Représentation schématique des trois composantes de la connectivité (d’après Calabrese & Fagan, 2004).
3
La connectivité réelle se réfère donc aux échanges d’individus entre populations locales spatialement séparées. Ces populations locales, évoluant sur des patchs d’habitats différents,
sont liées par des échanges de larves, juvéniles ou adultes résidant de façon permanente ou à
long terme dans la nouvelle population (Palumbi 2003). Un temps de résidence minimal est
requis afin de 1) distinguer les phénomènes de dispersion/migration des mouvements
temporaires associés aux activités de recherche de nourriture ou autres activités quotidiennes
et 2) contribuer au pool de gènes des nouvelles populations par transferts de gènes (Lowe &
Allendorf 2010). La connectivité réelle peut se mesurer à différentes échelles spatiotemporelles, notamment à l’échelle écologique (connectivité démographique) et à l’échelle évolutive (connectivité génétique) (Moilanen & Hanski 2001, Cowen et al. 2007). La
connectivité écologique (ou connectivité démographique) se traduit par des échanges
d’individus suffisants pour entrainer des impacts mesurables sur la dynamique des populations locales (Sale et al. 2010). Les mouvements d’individus se traduisent également
par des transferts de gènes permettant le maintien de l’homogénéité génétique entre populations locales, ce qui définit le concept de connectivité génétique (ou connectivité
évolutive) (Cowen et al. 2007, Cowen & Sponaugle 2009). Des échanges rares et faibles (un
individu par génération en moyenne) suffisent à maintenir l’homogénéité génétique entre deux populations locales (Hedgecock et al. 2007). Les échelles spatio-temporelles à
considérer dans le cas de la connectivité génétique sont bien supérieures par rapport aux
échelles à considérer dans le cas de la connectivité écologique. Ces différences d’échelles spatio-temporelles entre les deux types de connectivité sont expliquées plus en détails dans
l’exposé des méthodes de mesures de la connectivité (cf. 2. Méthodes de mesure de la
connectivité) (Taylor et al. 1993, Tischendorf & Fahring 2000, Moilanen & Hanski 2001).
1.2. La connectivité chez les poissons récifaux-lagonaires
Dans les écosystèmes marins, la connectivité réfère aux liens démographiques des
populations locales à travers la dispersion des individus aux stades larvaire, juvénile ou adulte
(Sale et al. 2010). Une des formes de connectivité considérée comme la plus importante pour
les organismes marins est représentée à travers le processus de dispersion des larves
planctoniques (Almany et al. 2007, Almany et al. 2009). Chez les poissons récifaux, la phase
larvaire pélagique est la première opportunité de dispersion des individus et ce phénomène de
connectivité a été particulièrement étudié (Thorrold et al. 2007, Cowen & Sponaugle 2009,
Jones et al. 2009). En effet, il existe de nombreuses études sur la durée de vie larvaire et le
4
comportement de celles-ci lors du retour vers les récifs (capacités natatoires et sensorielles)
(Fisher et al. 2000, Kingsford et al. 2002, Leis et al. 2007), l’estimation du phénomène d’autorecrutement (retour aux sites sources) ou de dispersion à grande échelle (Thorrold et al.
2007, Jones et al. 2009). D’après Jones et al (2009), les populations marines sont qualifiées
d’ouvertes, c'est-à-dire avec un fort potentiel de connectivité, lorsque la majorité du
recrutement est issu d’échanges entre populations locales. Au contraire, une connectivité limitée se traduit par des populations fermées c'est-à-dire présentant un fort taux
d’autorecrutement (Cowen et al. 2000).
De nombreuses espèces de poissons récifaux-lagonaires présentent des populations de
juvéniles et d’adultes généralement considérées comme sédentaires et territoriaux, connectées par une large phase de dispersion larvaire (Sale 1991). Cette idée reçue peut entrainer une
sous estimation des mouvements des adultes et juvéniles chez de nombreuses espèces. En
effet, la phase larvaire ne doit pas être considérée comme l’unique stade de vie impliqué dans la connectivité des communautés marines. Des études acoustiques révèlent qu’en réalité les mouvements des poissons sont bien plus importants que ce que l’on pensait (Chateau &
Wantiez 2009, Meyer et al. 2010, Claydon et al. 2012) et que les processus de dispersion sont
toujours possibles chez les populations adultes et juvéniles. D’ailleurs, de récentes études indiquent que le mouvement des adultes et juvéniles permet a de nombreuses espèces de
traverser les barrières de connectivité larvaire (Luiz et al. 2012) et que l’aire de distribution
géographique des poissons de récifs coralliens est plus fortement expliquée par les traits de
vie des adultes que par la durée de vie larvaire planctonique (Luiz et al. 2013).
1.2.1. La notion de « home range »
Le home range est défini comme la zone utilisée par un animal pour satisfaire les activités
quotidiennes telles que l’alimentation ou le repos, durant une partie ou la totalité de son existence (Grüss et al. 2011). L’étendue du home range doit pouvoir satisfaire les
mouvements entre la zone de refuge et la zone d’alimentation. Suivant l’écologie de l’espèce considérée, l’étendue de la zone de vie peut être de l’ordre de quelques mètres jusqu’à des dizaines de kilomètres et peut couvrir un à plusieurs habitats. Elle est plus importante pour les
espèces non territoriales que territoriales. Au sein de la même population, les patrons de
mouvements quotidiens peuvent être différents : certains individus présentent des
mouvements entre les zones distinctes de refuge et d’alimentation, d’autres individus restent relativement proches de leur refuge en permanence.
5
1.2.2. Les mouvements d’histoire de vie
Les migrations sont définies comme des mouvements généralement prévisibles dans le
temps, se produisant saisonnièrement ou à des instants particuliers du cycle de vie des
individus. L’étendue des migrations peut être de l’ordre de quelques centaines de mètres à des milliers de kilomètres. Les migrations d’individus juvéniles et/ou adultes peuvent se produire si les individus ont besoin d’atteindre des zones particulières de frai, de nurseries ou d’alimentation et/ou en réponse à des changements environnementaux (Grüss et al. 2011).
Les migrations ontogéniques se définissent par des changements d’habitat liés aux différents stades ontogéniques des individus (Fig. 1.2). La plupart des espèces de poissons
récifaux-lagonaires possèdent un cycle de vie complexe avec un stade larvaire pélagique puis
une phase benthique pour les juvéniles et adultes (Sale 1991). Chez certaines espèces de
poissons récifaux, les stades juvéniles et adultes occupent des habitats différents. Lorsque les
larves sont compétentes, elles rejoignent le milieu récifal et s’installent au niveau d’habitats côtiers peu profonds tels que les mangroves, les récifs ou les herbiers. A l’installation, elles se métamorphosent en juvéniles qui évoluent au sein de ces habitats qualifiés de zone de
nurseries car ils fournissent de nombreux avantages pour leur développement (refuges contre
les prédateurs, disponibilité accrue en ressources alimentaires adaptées, etc.) (Parrish 1989,
Beck et al. 2003). A mesure qu’ils grandissent, les individus vont rejoindre les populations adultes, ce processus correspond au recrutement et se traduit par l’intégration de nouveaux individus parmi les populations d’individus sexuellement matures (Sale 1991).
6
Figure 1.2 : Cycle biologique simplifié de Lutjanus fulviflamma, espèce à œufs pélagiques.
Les migrations liées à la reproduction sont associées à la recherche de partenaire ou de
lieux de reproduction favorables. Elles sont souvent décrites comme un phénomène de
mouvement des animaux de leur lieu de vie à un lieu de reproduction où ils se rassemblent en
grand nombre. Cependant, les échelles spatiales et temporelles des migrations de reproduction
peuvent s’étendre de distances quotidiennes inférieures à un kilomètre à des mouvements
annuels de plusieurs centaines de kilomètres selon l’espèce considérée (Sale 2002). Les
migrations courtes consistent généralement en des agrégations formées quotidiennement sur
des saisons de reproduction prolongées. Les migrations plus importantes sont généralement
caractérisées par des agrégations mensuelles formées sur des saisons de reproduction courtes
(Domeier & Colin 1997).
Ces liaisons fonctionnelles entre les habitats et leurs intensités sont essentielles pour le
bon fonctionnement des écosystèmes fragmentés du paysage récifo-lagonaire. Identifier et
quantifier les patrons de mouvements des juvéniles et des adultes est d’une importance cruciale pour la mise en place et l’efficacité des réseaux d’aires marines protégées (Grüss et
al. 2011).
7
1.3. Implications de la connectivité
Le déclin des récifs coralliens à travers le monde appelle au développement de stratégies
de protection de la résilience de ces écosystèmes (Pandolfi et al. 2003). Le mouvement est un
phénomène critique pour la persistance des populations animales et la résilience des
écosystèmes (Taylor et al. 1993) ; il est donc logique que la connectivité écologique soit
considérée comme un objectif clé dans la conservation des écosystèmes et la gestion des
ressources (Almany et al. 2009, Salas et al. 2010, Sale et al. 2010). L’idéal serait de pouvoir protéger l’ensemble des écosystèmes constitutifs du paysage récifo-lagonaire au sein des aires
marines protégées (AMP), afin de maximiser la probabilité de capturer l’ensemble des connections existantes. Cette solution n’est généralement pas envisageable et l’élaboration de réseaux d’AMP nécessite, au préalable, une quantification de la connectivité écologique entre
populations locales. En effet, les paramètres primordiaux intervenants dans le design des
réseaux (taille, nombre, surface totale, espacement et emplacement des réserves) ne peuvent
être décidés de façon adéquate sans une compréhension de la connectivité écologique
existante. Bien qu’il soit impossible d’estimer la connectivité pour l’ensemble des espèces marines présentes, il est important de pouvoir l’estimer pour une gamme d’espèces aux traits d’histoires de vie, degrés de menace et d’exploitation différents ceci dans le but d’étendre les
estimations obtenues à des espèces similaires à celles étudiées (Jones et al. 2007, McCook et
al. 2009).
Le maintien de la résilience ne se fait pas uniquement en préservant la connectivité
écologique, d’autres processus sont impliqués. L’identification et la protection des sites « sources » c'est-à-dire des sites producteurs nets d’émigrants (Jones et al. 2007, McCook et
al. 2009) est essentielle, ainsi que l’identification et la protection des habitats essentiels pour
les espèces présentant un cycle de vie complexe (migrations ontogéniques, de reproduction ou
trophiques) (Sale et al. 2010). Beck (2001) indique qu’il est crucial d’inclure les habitats qualifiés de nurseries dans les réseaux d’AMP dans le but de protéger à la fois les juvéniles
présents dans ces nurseries ainsi que les populations adultes renouvelées à travers le processus
de recrutement de ces juvéniles.
Pour l’étude de la connectivité des populations, il est nécessaire d’identifier des échelles
de temps pertinentes. Dans un objectif de conservation et de gestion des populations, l’étude de la connectivité à l’échelle écologique est plus pertinente qu’à l’échelle évolutive (Cowen et
al. 2007).
8
2. Méthodes de mesure de la connectivité
L’estimation de la connectivité des populations peut être réalisée grâce à différentes méthodes : la génétique, les techniques de marquage et la microchimie des otolithes
constituent des approches permettant l’étude de la dispersion et des mouvements des individus à différentes échelles spatiales et temporelles (Fig. 1.3). En effet, les méthodes de
mesure de la connectivité changent selon la cible étudiée et surtout les échelles spatiales et
temporelles choisies (Moilanen & Hanski 2006). Ainsi, l’estimation de la connectivité
écologique ne peut se réaliser à l’aide des mêmes méthodes utilisées pour quantifier la connectivité évolutive (Fig. 1.3).
Figure 1.3 : Echelles temporelles et spatiales correspondant aux différentes approches de
mesures de la dispersion des organismes récifaux-coralliens (modifiée de Jones et al. (2009)).
Les différentes approches permettant d’obtenir des informations sur la connectivité des populations nécessitent toutes l’emploi d’un marqueur. Ce dernier peut être d’origine
naturelle ou artificielle (Levin 2006).
9
2.1. Marqueurs artificiels
Les techniques de marquage artificiel font appel à une large gamme de marqueurs
différents, des marqueurs issus d’une technologie avancée (e.g. émetteurs satellites) à des
techniques basiques (e.g. marqueurs externes colorés) (Hastein et al. 2001). Les études avec
marqueurs externes simples nécessitent de suivre un protocole « capture-marquagerecapture » ne permettant pas de reconstruire en détails les mouvements effectués durant le
laps de temps entre la capture et la recapture. De plus, ces expériences deviennent rapidement
coûteuses car les taux de recapture sont en général très faibles (Hansen & Jacobsen 2003).
Parmi les techniques de marquage artificiel, le marquage acoustique permet d’obtenir des données remarquables sur le détail des déplacements des individus et ce, même pour les
espèces mobiles présentant des populations importantes (Kaunda-Arara & Rose 2004a,
Chateau & Wantiez 2009, Claydon et al. 2012). Cependant le suivi acoustique actif quantifie
le comportement de quelques individus sur des périodes de temps relativement courtes
(plusieurs semaines). Le suivi acoustique passif est possible à long terme (jusqu’à 2 ans) mais ne permet pas d’étudier les déplacements des individus tout au long de leur vie (Meyer et al.
2010).
Les otolithes des poissons sont des structures calcifiées idéales pour les techniques de
marquage chimique artificiel. Les composés fluorescents (e.g. tétracycline et calcéine) et
l’enrichissement en isotopes stables (e.g. baryum) sont des techniques utilisées pour marquer
les embryons ou les larves élevées en laboratoire avant la dispersion des sites natals (Jones et
al. 1999, Thorrold et al. 2002, Thorrold et al. 2007). Le marquage transgénérationnel par
enrichissement en isotopes stables est une méthode permettant un marquage en masse des
larves de poissons. Le procédé consiste à exposer des femelles gravides à une solution
enrichie en isotope stable d’un élément présent naturellement au sein des otolithes. Le marquage des otolithes des embryons se fait donc par transmission maternelle de l’élément
dont le rapport isotopique a été modifié artificiellement. Les éléments utilisés sont le
strontium ou le baryum via des solutions de chlorure de strontium ou de baryum (Thorrold et
al. 2002, Thorrold et al. 2006, Almany et al. 2007). Cette méthode, en plein essor, s’est principalement focalisée sur la connectivité des populations à travers la dispersion larvaire.
10
2.2. Marqueurs naturels
Les difficultés associées aux techniques de marquage artificiel nécessitant une recapture
des individus, ont conduit les chercheurs à s’intéresser aux marqueurs présents naturellement au sein des organismes.
Les méthodes génétiques indirectes (Fis, Fst) sont généralement inappropriées à
l’estimation de la connectivité démographique car elles fournissent des informations sur la connectivité génétique, définie comme étant le degré auquel les flux de gènes affectent les
processus évolutifs au sein des populations. Les méthodes génétiques seules apportent peu
d’informations sur la connectivité démographique, définie comme le degré auquel la croissance de la population et les taux vitaux sont affectés par la dispersion (Hedgecock et al.
2007, Lowe & Allendorf 2010). Les méthodes de mesure basées sur des approches
génétiques indirectes sont donc intégrées sur des échelles de temps multi-générationnelles,
bien plus longues que pour les autres méthodes de mesure comme les empreintes élémentaires
(Levin 2006).
Les méthodes génétiques directes ont émergées bien plus récemment. Ce sont des
techniques puissantes pour l’étude de la connectivité à des échelles de temps inférieures aux méthodes génétiques indirectes. Ces méthodes directes se focalisent sur l’assignement des individus à leur population d’origine (Manel et al. 2005) ou à leur parent (Jones et al. 2005).
Elles nécessitent l’établissement de marqueurs génétiques (e.g. microsatellites) pour l’espèce étudiée et l’identification de toutes les potentielles sources parentales. Cela représente un travail considérable qui n’est pas toujours logistiquement possible.
Aucune des méthodes précédentes ne permet de mesurer l’utilisation des habitats et la
connectivité au cours d’une vie entière, de la naissance à la mort d’un poisson. En revanche,
l’analyse de la microchimie des otolithes de poissons est une technique apportant un regard neuf sur les patrons continus de mouvements des poissons et permettant la mesure de la
connectivité à des échelles de temps écologiques (Patterson & Kingsford 2005, Levin 2006,
Cowen et al. 2007, Thorrold et al. 2007, Cowen & Sponaugle 2009).
11
3. Microchimie des otolithes
3.1. Généralités sur les otolithes
Les poissons Téléostéens possèdent, au sein de leurs oreilles internes, des structures
extracellulaires minéralisées nommées otolithes. Les otolithes font partie du système sensoriel
et sont impliqués dans les fonctions d’équilibration et de mécano-réception du poisson.
L’oreille interne des poissons téléostéens est composée d’un ensemble de trois canaux semicirculaires s’ouvrant à leur base sur une série de trois sacs otiques (Fig. 1.4). Ces trois sacs
otiques sont remplis d’endolymphe dans lequel baigne un otolithe. Les sacs otiques sont le
sacculus (saccule), l'utriculus (utricule) et la lagena qui contiennent respectivement la sagitta,
le lapillus et l'asteriscus. Les trois otolithes diffèrent par leur taille et leur forme, les sagittae
étant préférentiellement étudiées car ce sont les plus volumineuses (Fig. 1.4).
Malgré des parties communes entre les espèces (Fig. 1.5), la forme des otolithes présente
une
importante
variation
interspécifique.
Cette
variation
morphologique
permet
l’identification des espèces grâce à l’étude de la forme de leurs otolithes, ce qui se révèle être
utile pour la détermination des régimes alimentaires des oiseaux marins ou des hommes de
civilisations anciennes puisque les otolithes sont retrouvés dans les fouilles archéologiques
(Disspain et al. 2011, Mercier et al. 2012).
Figure 1.4 : Positionnement des otolithes au sein de l’oreille interne d’un téléostéen typique. (A) Vue dorsale de l’appareil vestibulaire ;; (B) Position des otolithes à l’intérieur du système
du labyrinthe (issue de Panfili et al. (2002)).
12
Les otolithes sont des formations composées principalement de carbonate de calcium
(CaCO3) sous la forme de cristaux d’aragonite précipités dans une matrice protéique
(Campana 1999, Panfili et al. 2002). Ils s’accroissent par additions journalières de couches concentriques de CaCO3 et de protéines (en majorité de l’otoline) à partir du noyau central (primordium) (Fig. 1.5). Ces dépôts successifs sont à l’origine de la formation de micro et macrostructures utilisées dans les études d’âge et de croissance des individus (Panfili et al.
2002). Les microstructures correspondent aux dépôts journaliers et forment une succession de
bandes concentriques sombres et claires permettant d’estimer l’âge en jours des jeunes poissons (larves et juvéniles). Les macrostructures sont constituées d’une succession de zones
claires et sombres correspondant aux différentes saisons vécues par les individus. Elles
peuvent être interprétées en marques annuelles et permettent d’évaluer l’âge des poissons en années. De même, certains évènements marquant de l’histoire de vie des poissons tels que la
métamorphose ou la reproduction forment des structures particulières observables sur les
otolithes (Campana & Neilson 1985).
Figure 1.5 : Morphologie d’une sagitta typique, face interne et externe (A) et détails des trois
plans d’orientations d’une sagitta typique (B) (issue de Panfili et al. (2002)).
Ces multiples caractéristiques des otolithes sont utilisés dans différents domaines : étude
de l’âge et de la croissance (Choat & Robertson 2001, Choat et al. 2003, Labonne et al. 2008,
Vigliola & Meekan 2009), études halieutiques (traits démographiques, dynamique et
évaluation des stocks, aide à la gestion des pêches) (Clausen et al. 2007), étude des stades
marquants de l’histoire de vie des individus (durée de phase larvaire, métamorphose, 13
recrutement, reproduction) (Lecomte-Finiger 1992, Searcy & Sponaugle 2000, Grandcourt et
al. 2006) et études taxonomiques (morphométrie des otolithes) (Campana & Casselman 1993,
Pothin et al. 2006, Barnes & Gillanders 2013).
Depuis le début des années quatre-vingt, la composition chimique des otolithes est de
plus en plus étudiée notamment grâce aux progrès technologiques des appareils de mesures
employés en chimie analytique et géochimie (e.g. ICP-MS solution, laser, etc.) (Campana &
Neilson 1985) et pour toutes les perspectives qu’elle ouvre en terme d’évaluation de la connectivité entre populations.
3.2. Microchimie des otolithes
La microchimie des otolithes repose sur une hypothèse simple : à travers les dépôts
journaliers, des éléments chimiques provenant de l’environnement dans lequel évolue le poisson sont incorporés dans les otolithes. L’eau environnante est considérée comme la source dominante des éléments. Jusqu’ici 50 éléments ont été détectés à différentes concentrations au
sein des otolithes. Des éléments majeurs présents en fortes concentrations (Ca, C, O et N), des
éléments mineurs (> 100 ppm : Cl, S, Mg, Na, P, Sr et K) et des éléments traces (< 100 ppm :
Ag, Al, As, B, Bi, Ba, Br, Cd, Ce, Co, Cr, Ce, Cu, Dy, Er, Fe, Gd, Hg, Ho, La, Li, Lu, Mn,
Nd, Ni, Pb, Pr, Rb, Sc, Se, Si, Sm, Tb, Tm, U, V, Y, Yb et Zn) (Sturrock et al. 2012).
L’incorporation des éléments se fait en plusieurs étapes car les otolithes ne sont pas en contact
direct avec l’environnement extérieur. En effet, les éléments doivent franchir différentes barrières biologiques avant d’être incorporés : ils passent de l’eau environnante au plasma sanguin via les branchies ou les intestins, traversent les membranes de l’oreille interne vers l’endolymphe, pour finalement atteindre la surface de l’otolithe (Fig. 1.6). Les éléments sujets
à de fortes régulations physiologiques (e.g. Na, K, S, P, Cl) sont d’un intérêt limité en tant que
traceurs de l’environnement (Campana 1999, Sturrock et al. 2012). A l’inverse, les éléments qui ne sont pas soumis à de fortes régulations physiologiques présentent potentiellement une
concentration dans les otolithes à l’image de leur concentration dans l’eau environnante.
La méthode de microchimie repose sur les deux propriétés clés des otolithes 1) une
croissance continue au cours de la vie des poissons et 2) une nature métaboliquement inerte
assurant l’inaltération de la composition chimique à travers le temps. Ces propriétés ont 14
permis de qualifier les otolithes d’enregistreurs permanent de l’environnement, archives de l’histoire de vie environnementale des poissons (Campana 1999).
Figure 1.6 : Voies d’incorporation des éléments, barrières physiologiques entre l’eau et les otolithes et estimations des taux de transfert pour certains éléments à chaque barrière
physiologique (issue de Campana (1999)).
La composition chimique des otolithes constitue un traceur puissant des masses d’eaux traversées par les poissons. Elle a permis au cours de plusieurs études de retracer l’histoire environnementale des individus (McCulloch et al. 2005) et de discriminer spatialement, et à
différentes échelles, des populations ou groupes de poissons (Kingsford & Gillanders 2000,
Gillanders et al. 2001, Rooker et al. 2003, Chittaro et al. 2004, Patterson et al. 2004).
Cependant, certaines études n’ont pas pu mettre en évidence une variabilité spatiale suffisante
des signatures élémentaires pour discriminer différentes zones géographiques (Chittaro et al.
2005 , Patterson & Kingsford 2005). Cette absence de discrimination serait expliquée par un
manque de variabilité spatiale au niveau de la chimie de l’environnement étudié. En effet, ces études ont été conduites dans des systèmes côtiers où les apports terrigènes limités entrainent
une chimie de l’environnement homogène. L’hétérogénéité spatiale des signatures microchimiques et les concentrations élevées de certains éléments dans les otolithes seraient
favorisées par des apports naturels d’éléments traces (e.g. géologie locale, apports fluviaux,
etc.) ou par des apports liés aux activités anthropiques (pollution, rejet des eaux usées,
exploitation minière, etc.) (Dove & Kingsford 1998, Morales-Nin et al. 2007).
15
4. Contexte en Nouvelle-Calédonie
Les lagons de Nouvelle-Calédonie se présentent comme un terrain particulièrement
intéressant pour l’étude de la connectivité par microchimie. Inscrits au patrimoine mondial de l’UNESCO depuis 2008, ils constituent un des trois systèmes récifaux les plus importants au
monde (Andréfouët et al. 2009). Ce paysage récifaux-lagonaire rassemble une grande
diversité d’écosystèmes contrastés : mangroves, herbiers et algueraies, fonds meubles, récifs
coralliens (Kulbicki & Rivaton 1997, Andréfouët & Torres-Pulliza 2004). Leur distribution,
plus ou moins fragmentée, forme une mosaïque d’habitats abritant de multiples assemblages
de poissons interconnectés (Parrish 1989, Wantiez 2008). De part la superficie et la
morphologie de la Nouvelle-Calédonie et de ses lagons, les écosystèmes lagonaires sont sous
influences terrigène et océanique contrastées à l’origine d’une microchimie de l’environnement hétérogène. Cette hétérogénéité est soupçonnée de se traduire par des
gradients environnementaux marqués, notamment un gradient côte-large.
La Nouvelle-Calédonie présente également une géologie spécifique, extrêmement riche
en différents minerais. Elle constitue la troisième réserve mondiale de nickel dont
l’exploitation occupe une place majeure dans l’économie locale. D’autres minerais tels que le chrome, cobalt, fer, cuivre, plomb sont présents dans le sol calédonien (Ambatsian et al.
1997). Certains de ces gisements (nickel, chrome, cobalt et fer) sont associés aux massifs
ultrabasiques (ou ultramafiques) couvrant environ le tiers de la superficie de la Grande Terre,
principalement dans le sud et sur la côte ouest (Paris 1981, L'Huillier et al. 2010). Sur le
territoire, les activités de prospection et d’exploitation de minerais ont commencé à la fin du
19ème siècle et les aménagements du territoire associés (dévégétalisation, dragages, mines à
ciel ouvert, etc.) constituent les principales sources de perturbations environnementales en
Nouvelle-Calédonie (Labrosse et al. 2000, Fichez et al. 2008). Elles seraient susceptibles de
venir enrichir les signatures microchimiques en éléments tels que le nickel, le chrome ou le
cobalt. De ce fait, la microchimie de l’environnement lagonaire pourrait dévoiler des signatures multi-élémentaires spécifiques aux sites et habitats étudiés (Fernandez et al. 2006).
La morphologie du lagon, le contexte géologique singulier additionnés à des études
antérieures prometteuses (Labonne et al. 2008, Sigura 2009), indiquent que l’outil
microchimique aurait un potentiel informatif et discriminant élevé en Nouvelle-Calédonie.
16
5. Objectifs de la thèse
La thèse étant construite sous forme d’articles scientifiques, les matériel et méthodes sont
détaillés au sein de chaque article/chapitre. Cependant, il est apparu indispensable de décrire
la méthodologie générale en français avant d’aborder les différentes thématiques. Une fois la
méthodologie générale exposée dans le chapitre 2, ce travail de recherche se divise en deux
axes principaux que sont 1) le développement méthodologique de l’outil microchimique en
Nouvelle-Calédonie et 2) son utilisation dans l’étude de la connectivité. Le premier axe, constitué des chapitres 3 et 4, est composé d’un travail de détermination et de précision des signatures microchimiques des otolithes et de l’environnement. Le
chapitre 3 s’intéressera uniquement à la caractérisation de la microchimie des otolithes en
répondant aux problématiques suivantes : Quels éléments caractérisent les variations
spatiales à l’échelle des sites et habitats en Nouvelle-Calédonie ? Quelle est la précision
des signatures microchimiques mono-spécifiques et comment varie-t-elle en fonction de
différentes espèces à traits de vie contrastés ? Quelle est la perte de précision lors de
l’utilisation de signatures microchimiques multi-spécifiques comme proxy des signatures
mono-spécifiques ? Ce chapitre utilisera les données de microchimie issues de l’analyse du bord des otolithes traitées indépendamment pour plusieurs espèces cibles (signatures monospécifiques) et traitées en regroupant toutes les espèces indépendamment de leur identité ou
en groupes taxonomiquement ou écologiquement proches (signatures multi-spécifiques).
Le chapitre 4, se penchera sur l’étude des relations entre la microchimie de
l’environnement et des otolithes, en répondant aux questions suivantes : Quels éléments issus
de l’environnement caractérisent les variations spatiales à l’échelle des sites et habitats
en Nouvelle-Calédonie ? Quelle est la précision des signatures chimiques issues de
l’environnement et comment varie-t-elle
en
fonction
du
compartiment
environnemental ? Quelles sont les relations entre la microchimie des otolithes de
plusieurs espèces, celles de l’environnement et celles des paramètres physico-chimiques
de la masse d’eau ? Ce chapitre utilisera les données de température et de salinité de l’eau ainsi que les données de microchimie issues des analyses de l’eau de mer, des sédiments de surface et du bord des otolithes pour différentes espèces traitées indépendamment (signatures
mono-spécifiques) et pour l’ensemble des espèces (signature multi-spécifique).
17
Le volet méthodologique permettra de poser les bases microchimiques pour le second
axe : l’observation et la quantification de la connectivité entre différents habitats du lagon à
travers l’étude de la daurade à tâche noire, Lutjanus fulviflamma (Forsskål, 1775). Cet axe est
composé de deux chapitres. Le chapitre 5 focalisera sur l’estimation du rôle des mangroves de Nouvelle-Calédonie comme sites juvéniles pour L. fulviflamma à travers les questions
suivantes : Quelle proportion d’individus adultes ont évolué au sein d’une mangrove en tant que juvénile ? La présence de mangrove est-elle un déterminant primordial à la
distribution de l’espèce à l’échelle de l’archipel de la Nouvelle-Calédonie ? Ce chapitre
utilisera les données microchimiques du bord des otolithes de L. fulviflamma pour les
mangroves et les récifs de l’ensemble de l’île en vue de construire les fonctions
discriminantes de ces deux habitats. Une fois construites, les fonctions discriminantes seront
appliquées à la partie des données microchimiques des sections transversales d’otolithes d’individus adultes correspondant à leur phase juvénile. Elles permettront de prédire les habitats utilisés au cours de la période juvénile et de définir le statut des mangroves
calédoniennes comme des habitats juvéniles accessoires, essentiels ou obligatoires pour L.
fulviflamma. A cela s’ajouteront des données de comptages sous-marins (UVC – Underwater
Visual Census) et une cartographie des mangroves en Nouvelle-Calédonie qui permettra de
mettre en relation la surface des mangroves avec la présence et l’abondance de L. fulviflamma
dans les récifs adjacents.
Le chapitre 6 se focalisera sur la reconstruction de l’histoire environnementale des L.
fulviflamma via les questions suivantes : Quel est le laps de temps d’utilisation des mangroves par les juvéniles et varie-t-il selon les individus ou les sites ? Quelle transition
est la plus fréquente entre les habitats juvéniles et adultes ? Quels sont les patterns de
mouvement des adultes ? Ce chapitre sera constitué des données microchimiques du bord
des otolithes de L. fulviflamma pour les mangroves et les récifs frangeants, intermédiaires et
barrières de quatre sites de la côte Ouest de l’île en vue de construire les fonctions discriminantes de ces habitats. Une fois les fonctions construites, elles seront utilisées pour
prédire les habitats fréquentés au cours de la vie des individus via la classification des
données des transects microchimiques réalisés sur les sections transversales d’otolithes de L.
fulviflamma adultes.
18
Chapitre 2 : Méthodologie générale
19
20
1. Sites d’étude
1.1. La Nouvelle-Calédonie
La Nouvelle-Calédonie est une collectivité française d’outre mer située dans le Pacifique Sud-Ouest à 1500 km à l’est de l’Australie, entre les latitudes 19°S et 23°S et les longitudes
163°E et 168°E. L’archipel d’une superficie de 19 954 km2 est constitué d’une île principale
(Grande-Terre), de trois groupes d’îles (Loyautés, Ile des Pins et Bélep) et de quelques îles inhabitées (Entrecasteaux, Chesterfield, etc.). La Grande Terre présente une superficie de
17 000 km2, allongée sur environ 400 km selon une direction nord-ouest/sud-est et d’une largeur de 60 km au maximum (Paris 1981, Andréfouët et al. 2009) (Fig. 2.1).
Le lagon de la Grande Terre est constitué d’un récif barrière de 1744 km de long
entrecoupé de passes (Andréfouët & Torres-Pulliza 2004). La distance entre la Grande Terre
et le récif barrière varie de 1 à 65 km. Cette barrière délimite un lagon dont la superficie
s’élève à 31 336 km2 et dont la profondeur atteint 40 m en certains endroits (Andréfouët et al.
2009). Depuis 2008, quatre zones du lagon de la Grande Terre sont inscrites sur la Liste du
Patrimoine Mondial de l’Unesco (le Grand Lagon Sud, la Zone Côtière Ouest, la Zone Côtière
Nord et Est et le Grand Lagon Nord).
Figure 2.1 : Situation géographique de la Nouvelle-Calédonie dans le Pacifique Sud-Ouest.
21
1.2. Sites d’échantillonnages
Onze sites répartis autour de la Grande Terre ont été choisis en fonction de différents
critères : le type et la présence de certains habitats, les activités minières voisines, l’existence d’études scientifiques antérieures, le statut et les projets de gestions (e.g. inscription au
Patrimoine Mondial de l’Unesco ou classement en AMP) et l’accessibilité (Tab. 2.1 et Fig.
2.2). Pour les onze sites, l’échantillonnage est effectué dans deux habitats : la mangrove et le
récif barrière interne.
Parmi ces onze sites, quatre sites ateliers situés sur la Côte Ouest de l’île ont été
identifiés : deux sites potentiellement « impactés » par des projets miniers majeurs : les
régions de Voh (Gatope) (projet Koniambo Nickel-SAS) et Prony (projet Vale Inco Nouvelle
Calédonie) et deux sites « non impactés » : Ouano (Patrimoine mondial) et St Vincent. Dans
ces quatre sites, quatre habitats sont échantillonnés selon un gradient côte-large: la mangrove,
les récifs coralliens frangeants, intermédiaires et barrières internes.
Tableau 2.1: Les onze sites d’échantillonnages (* Sites ateliers; M : mangrove ; RF : récif
frangeant ; RI : récif intermédiaire ; RB : récif barrière interne).
N°
Sites
Caractéristiques
n stations
Habitats
1
Nouméa
Usine pyrométallurgique de Doniambo (SLN : Société Le Nickel)
Pollution urbaine
2
M / RB
2*
St Vincent
Site non-minier dans sa partie échantillonnée
Etudes antérieures
4
M / RF / RI / RB
3*
Ouano
Site non-minier
Inscription au Patrimoine Mondial de l’Unesco
4
M / RF / RI / RB
4
Népoui
Exploitation minière de Népoui-Kopéto (SLN)
2
M / RB
5*
Gatope
Projet minier Koniambo Nickel SAS (KNS)
Développement de la zone urbaine Voh-Koné-Pouembout
4
M / RF / RI / RB
6
Golonne
Site non-minier
2
M / RB
7
Amos
Site non-minier
2
M / RB
8
Tchambouen
ne
Site non-minier
2
M / RB
9
Paama
Site non-minier
2
M / RB
10
Port Bouquet
Exploitation minière de Thio (SLN)
2
M / RB
11*
Prony
Projet minier Vale Inco Nouvelle-Calédonie
Présences de cheminées hydrothermales
4
M / RF / RI / RB
Nombre total de stations
22
30
Figure 2.2 : (A) Sites d’étude : les cercles jaunes et rouges désignent respectivement les sites
à 2 et 4 habitats échantillonnés. (B et C) Détails des habitats échantillonnés au sein des sites à
2 et 4 habitats échantillonnés. (Base de la carte : Millennium Coral Reef Mapping
(Andréfouët & Torres-Pulliza 2004) ; réalisation : M. A. Hamel).
1.3. Description des sites ateliers
La baie de Saint-Vincent est située à 40 km au nord de Nouméa. D’une superficie de 172
km2, c’est la baie la plus vaste de la côte ouest (Testau & Conand 1983) qui englobe de
nombreux îlots (Fig. 2.3). Les arrivées d’eau douce proviennent de trois rivières : la Tontouta,
la Tamoa et la Ouenghi. La zone de mangrove St Vincent/Ouenghi, définie comme un sous
ensemble géographique par Thollot (1992b), couvre une surface égale à 2.6 km2. Il existe
deux exploitations minières en activité au niveau du bassin versant de la Tontouta : Tomo
(SMGM) et Opoué (SLN/SMGM). Le point de chargement du minerai issu des deux sites
miniers se situe au sein de la baie de Saint-Vincent, à l’embouchure de la Tontouta.
Cependant, les sites échantillonnés étant éloignés du point de chargement, Saint-Vincent est
considéré comme non impacté.
23
Le site de Ouano se situe au niveau de la baie de Chambeyron, à plus de 100 km au nord
de Nouméa (Fig. 2.3). Ce site est inscrit au patrimoine mondial de l’Unesco, au sein de la Zone Côtière Ouest et présente des zones protégées juxtaposées à des aires non protégées. La
surface de mangrove de la zone Chambeyron/Ouano est égale à 6.5 km2.
La ville de Voh, située à 300 km de Nouméa au nord-ouest de la Grande Terre, est proche
du site de Gatope (Fig. 2.3). Ce site se situe au niveau sud de la baie de Chasseloup, d’une superficie de 75 km2 et au sein de laquelle il existe deux arrivées principales d’eau douce issues de deux bras d’une même rivière, la Témala (Testau & Conand 1983). Les premières
extractions de minerai se situent sur le massif du Koniambo et ont commencé à la fin du 19 ème
siècle. L’exploitation minière de Koniambo Nickel SAS, dont les travaux ont commencé en 2008, se situe au niveau de la baie de Vavouto, voisine de Chasseloup (Grenon & Simard
2012). Elle est constituée, entre autres, d’une mine à ciel ouvert, d’une centrale électrique et
d’une zone portuaire (chenal d’accès, quai de déchargement, etc.). La production de l’usine a
commencé en 2013 et devrait atteindre sa pleine activité en 2015.
La baie de Prony, située au sud de la Grande Terre, est une baie fermée d’une superficie
totale de 49 km2 (Testau & Conand 1983) (Fig. 2.3). Quatre rivières principales constituent les
apports d’eaux douces : Rivière du Kaoris, Rivière Bleue, Rivière du Carénage et Rivière
Kadji. C’est une baie qui présente une transition entre un environnement sous forte influence
terrigène et un environnement océanique sur une courte distance, soupçonnée de présenter une
microchimie de l’environnement singulière due à la présence de cheminées hydrothermales et
à la présence du projet minier Vale Inco Nouvelle-Calédonie (anciennement Goro Nickel)
(Bonvallot & Lardy 2012). Les premiers travaux sur le site ont commencé en 1998 et incluent,
entre autres, la construction de l’usine, de la centrale thermique et du port. La mine est une exploitation à ciel ouvert en gradins. La production a progressivement démarré depuis 2010
mais, à l’heure actuelle, n’a pas atteint sa pleine capacité.
24
Figure 2.3 : Détails des quatre sites ateliers: (A) Saint Vincent, (B) Ouano, (C) Gatope et (D)
Prony.
25
2. Plan d’échantillonnage
2.1. Campagnes d’échantillonnages
Trois campagnes d’échantillonnages ont été réalisées, une durant l’été austral (novembre à février 2009 ; Sigura 2009) et deux durant l’hiver austral (juin à septembre 2010 et 2011)
afin d’intégrer de la variabilité saisonnière et interannuelle dans la caractérisation des
signatures microchimiques. Lors de ces campagnes, les échantillonnages ne sont pas du même
niveau et les objectifs attenants sont détaillés dans le tableau (Tab. 2.2).
Tableau 2.2 : Campagnes d’échantillonnages réalisées et objectifs scientifiques reliés.
Périodes
Sites
Prélèvements Analyses
Chapitres Objectifs
Eté
2009
2 sites
Toutes espèces Bords otolithes entiers
3, 4, 5, 6
Variabilité spatiale des signatures
Elaboration des fonctions discriminantes
Hiver
2010
11 sites
3, 4, 5, 6
5, 6
Détermination de l’habitat juvénile
(n = 44 esp)
Toutes espèces Bords otolithes entiers
(n = 26 esp)
Transects
microchimiques sur
coupes transversales de L.
fulviflamma
Hiver
2011
4 sites
Analyse détaillée des migrations
Echantillons
d’eau et de sédiment
Analyses microchimiques 4
des échantillons d’eau et de sédiments
Corrélation microchimie
environnement/otolithes
Cibles
(n = 4 esp)
Bords otolithes entiers
3, 4, 5, 6
Deux niveaux d’échantillonnage sont réalisés suivant les axes de recherche de la thèse. Le premier niveau est composé d’un échantillonnage multi-spécifique large. Le second niveau
est réduit à quatre espèces cibles, présentes dans le premier niveau.
2.2. Analyses multi-spécifiques
L’étude de la variabilité interspécifique des signatures microchimiques et la recherche des
meilleures signatures multi-spécifiques requièrent un échantillonnage composé de plusieurs
espèces. De même, la corrélation entre la microchimie de l’environnement et des otolithes étant susceptible de varier entre les espèces (Gillanders & Kingsford 2003, Swearer et al.
26
2003, Hamer & Jenkins 2007), son étude nécessite un échantillonnage multi-spécifique. Pour
ces études, la microchimie à analyser est celle la plus récemment incorporée, révélée par
l’analyse du bord de l’otolithe entier. Une seule analyse par otolithe (i.e. par individu) étant
réalisée, le nombre d’individus et d’espèces inclus dans les analyses peut être élevé. Les échantillonnages multi-spécifiques larges ont été réalisés lors des deux premières campagnes
(Eté 2009 et Hiver 2010). Une liste d’espèces à collecter a été constituée sur la base de
différents critères même si au final les aléas de la pêche ont défini les échantillons
disponibles :
o Plusieurs espèces pour chacune des familles majeures (Acanthuridae, Chaetodontidae,
Lutjanidae, Lethrinidae, Pomacentridae, Scaridae, Serranidae, Siganidae) afin de
pouvoir former des groupes taxonomiques.
o Plusieurs espèces pour chacun des quatre groupes trophiques majeurs (piscivore,
carnivore, herbivore-détritivore, planctonivore) définis selon la littérature (Kulbicki et
al. 2005a, Kulbicki 2006, Chabanet et al. 2010, Guillemot et al. 2011) afin de pouvoir
former des groupes fonctionnels (i.e. groupes trophiques).
Les regroupements taxonomique ou fonctionnel permettent de déterminer si des
signatures multi-spécifiques regroupant des espèces proches (issues de la même famille ou
ayant un régime alimentaire similaire) peuvent être utilisées comme de meilleurs proxy des
signatures mono-spécifiques que des signatures multi-spécifiques globales, regroupant toutes
les espèces analysées.
2.3. Espèces cibles
Parmi les espèces collectées lors des échantillonnages multi-spécifiques de 2009 et 2010,
quatre espèces modèles ont été collectées à nouveau lors de l’échantillonnage hivernal 2011. Le choix s’est porté sur des espèces avec des traits de vie contrastés (niveau de mobilité,
mode de reproduction, occupation des habitats, gamme de taille, etc.) détaillés ci-après.
Lutjanus fulviflamma (Forsskål, 1775) appartient à la famille des Lutjanidae (Fig. 2.4).
Cette espèce présente une distribution étendue de l’Afrique de l’Est aux Samoa et du nord de l’Australie aux îles Ryukyu (Japon) (Carpenter & Niem 2001a, Allen et al. 2003). Les L.
fulviflamma arborent une tâche noire caractéristique au niveau de la ligne latérale d’où la désignation commune en Nouvelle-Calédonie de « daurade à tâche noire » (Laboute &
Grandperrin 2009). C’est une espèce carnivore dont le régime alimentaire est majoritairement
27
constitué de crustacés de toutes tailles et de poissons (Kulbicki et al. 2005a). D’après Loubens (1980), les L. fulviflamma présentent une période de reproduction comprise entre septembre et
février en Nouvelle-Calédonie alors qu’ils sont caractérisés par une période de reproduction plus large (de août à mars) selon la FAO (Carpenter & Niem 2001a). Les individus adultes
mesurent en moyenne 25 cm et la longueur totale maximale observée est de 35 cm (Carpenter
& Niem 2001a). Cette espèce mobile évolue généralement en petits groupes constitués de 3 à
25 individus (Kulbicki 2006, Guillemot et al. 2011) cependant quelques individus peuvent
être trouvés seuls, cachés au sein des anfractuosités des récifs. L. fulviflamma présente un
cycle de vie complexe et il fréquente différents habitats au cours de son développement. Les
adultes se trouvent sur les récifs coralliens à des profondeurs comprises entre 3 et 35m et
peuvent aussi évoluer en eaux saumâtres, à la limite des eaux douces. En Nouvelle Calédonie,
les juvéniles sont observés dans les zones côtières peu profondes telles que les mangroves
(Thollot 1992a).
Ctenochaetus striatus (Quoy & Gaimard, 1825) appartient à la famille des Acanthuridae
(Fig. 2.4). C’est une espèce abondante dont la distribution s’étend sur tout l’Indo-Pacifique
(Allen et al. 2003). Elle est herbivore et détritivore, se nourrissant à la fois du film algal
présent à la surface des substrats et de détritus (Carpenter & Niem 2001b, Laboute &
Grandperrin 2009). C’est une espèce grégaire vivant en groupe de taille moyenne (25 à 50
individus) (Guillemot et al. 2011). C. striatus est une espèce ubiquiste récifale, habitante de
récifs coralliens des lagons protégés aux zones exposés. Elle est qualifiée de sédentaire avec
un home range restreint à une centaine de m2, mais est soupçonnée être plus mobile (Claydon
et al. 2012). Elle ne présente pas d’intérêt commercial en Nouvelle-Calédonie car elle est
responsable de graves intoxications ciguatériques (Laboute & Grandperrin 2009). C. striatus
est une des espèces les plus abondantes du lagon de Nouvelle-Calédonie.
Dascyllus aruanus (Linnaeus, 1758) appartient à la famille des Pomacentridae (Fig. 2.4).
Elle présente un régime alimentaire de type planctonivore et évolue en groupe de taille
moyenne (25 à 50 individus) (Kulbicki 2006, Guillemot et al. 2011). C’est une espèce présente dans l’ensemble du lagon qui est sédentaire, inféodée à des coraux branchus (type
Acropora et Pocillopora) (Carpenter & Niem 2001a, Allen et al. 2003). C’est une espèce à œufs benthiques c’est-à-dire que lors de la ponte les œufs ne sont pas relâchés dans la masse d’eau mais collés au corail. Contrairement aux deux espèces précédentes, D. aruanus est une
28
espèce territoriale dont le home range se limite à quelques mètres carrés qu’elle défend contre les intrus (Sale 1971, Guillemot et al. 2011). Elle constitue le candidat idéal pour le rôle
d’espèce contrôle car, pour un lieu donné, les variations de la microchimie des otolithes ne
seront pas induites par le mouvement des individus mais uniquement par les variations
temporelles de la microchimie des eaux environnantes.
Siganus lineatus (Valenciennes, 1835) appartient à la famille des Siganidae et est présent
dans l’Indo-Pacifique Ouest (Fig. 2.4). C’est une espèce macro-herbivore, très mobile à l’âge adulte, grégaire se déplaçant en groupes composés de nombreux individus (> 50) (Guillemot
et al. 2011). Les juvéniles évoluent au sein d’habitats côtiers peu profonds tels que les mangroves et herbiers alors que les adultes fréquentent les récifs coralliens et substrats
rocheux peu profonds des eaux protégées lagonaires. Les individus adultes mesurent en
moyenne 25 cm et la longueur totale maximale observée est d’environ 45 cm (Carpenter &
Niem 2001b).
Figure 2. 4 : Photo des espèces modèles : (A) Lutjanus fulviflamma, (B) Dascyllus aruanus,
(C) Ctenochaetus striatus et (D) Siganus lineatus (Crédits photos Paillon, C. et Randall, J.E.).
3. Techniques de prélèvement
3.1. Mangroves
La méthode de prélèvement consiste à poser, à marée haute, un ensemble de filets
maillants d’une longueur totale de 250 m le long des palétuviers et de récupérer les individus
à marée basse (Fig. 2.5). Afin d’obtenir des espèces de tailles diverses ainsi que différents stades ontogéniques (juvénile et adultes), l’ensemble est composé de six filets de tailles de maille différentes : quatre filets de 50 m de longueur, d’1,5 m de chute et de 45 mm de taille
de maille carrée auxquels s’ajoute deux filets de 25 m de long, d’1,5 m de chute et de 25 mm
29
de taille de maille carrée. Afin d’échantillonner les juvéniles dissimulés entre les racines de palétuviers, un pied isolé est entouré à marée haute de deux filets de 25 m de long, 1,5 m de
chute et 7 mm de taille de maille. Une substance épaisse issue d’un mélange d’1 kg de poudre de roténone et d’eau de mer est versée à marée basse au niveau de la zone entourée. La roténone est une molécule toxique provoquant un empoisonnement des poissons présents dans
la zone ciblée. Un laps de temps d’environ 15 minutes est nécessaire avant d’observer les premiers poissons remonter à la surface. Tous les individus empoisonnés sont récoltés à l’aide d’épuisettes. 3.2. Récifs
Les prélèvements sont réalisés en chasse sous marine, principalement au fusil harpon.
Pour les poissons de petite taille (e.g. Chaetodontidae et Pomacentridae), la collecte est
réalisée à l’aide de fléchettes fines propulsées par un élastique. Dascyllus aruanus est
récupéré grâce à l’utilisation d’un mélange d’eugénol et d’alcool (Fig. 2.5). L’eugénol est un composé anesthésiant provoquant, à dose adéquate, un étourdissement des individus et
permettant de les récupérer facilement à l’épuisette (Soto & Burhanuddin 1995, Munday &
Wilson 1997).
Sur place, les échantillons récoltés sont stockés dans une glacière avant d’être congelés en chambre froide (-20°C) au laboratoire.
Figure 2.5 : Illustration des méthodes de prélèvements ; (A) au niveau des mangroves à
l’aide de filets maillants et (B) au niveau des récifs au fusil harpon et à l’eugénol.
30
3.3. Echantillons de l’environnement
Lors de la campagne hivernale de 2010, trois prélèvements d’eau ont été réalisés sur
chaque site et pour chaque habitat. En mangrove, les échantillons sont prélevés à marée basse,
au niveau de la couche d’eau de surface. Pour les récifs, les prélèvements sont aussi effectués au niveau de la couche de surface, sans tenir compte de la marée. Chaque échantillon est
prélevé avec une seringue de 40ml équipée d’un filtre à membrane en polyesthersulfone
(membrane PES de porosité 0,45µm) et uniquement 10ml de l’échantillon filtré est conservé
pour analyse dans des tubes en polypropylène de 15 ml. Les échantillons sont ensuite fixés
par ajout de 0,2 ml d’HNO3 Suprapur® à 2% afin d’éviter le développement des organismes vivants présents dans l’eau. Ils sont conservés au laboratoire, à l’abri de la lumière.
Durant la même campagne, trois prélèvements d’1kg de sédiments de surface sont
réalisés à chaque station. En mangrove, les prélèvements sont réalisés à marée basse et
espacés de 10 mètres. Au niveau des récifs, ils sont effectués en apnée, sans tenir compte de la
marée et également espacés de 10 mètres. Les échantillons sont stockés dans une glacière sur
le terrain puis congelés dans une chambre froide (-20°C) du laboratoire.
Les paramètres environnementaux, température et salinité, sont également mesurés à trois
reprises au niveau de la couche d’eau de surface pour chaque habitat échantillonné. La température est mesurée grâce à une sonde thermomètre Testo® d’une précision de 0,1°C. Un réfractomètre automatique Atago® d’une précision de 0.1‰ est utilisé pour les mesures de
salinité.
4. Protocole de préparation des échantillons
L’ensemble de l’étude se déroule en suivant des précautions de non contamination
métallique nécessaires pour les analyses de microchimie. Pour les extractions des otolithes,
des pinces céramiques sont utilisées et la totalité du matériel employé est préalablement
nettoyé à l’acide nitrique (HNO3, 2%), rincé à l’eau ultra-pure et séché sous hotte à flux
laminaire en conditions HEPA 100. De même, le matériel est rincé à l’eau ultra-pure avant et
après dissection de chaque individu.
Les individus collectés sont mesurés au mm près selon la longueur à la fourche (LF) puis
sont pesés au mg près. Les otolithes sont ensuite extraits puis plongés dans deux bains
31
consécutifs d’eau ultra-pure afin d’éliminer les éventuels tissus résiduels. Ils sont ensuite
placés à sécher sous une hotte à flux laminaire (conditions HEPA 100), et conservés par paire
dans des microtubes labellisés.
4.1. Otolithes in toto
Les analyses du bord des otolithes permettent de révéler la microchimie la plus
récemment incorporée au sein des otolithes, caractéristique du lieu d’échantillonnage. Ces analyses permettent la détermination des signatures mono et multi-spécifiques, l’élaboration des fonctions discriminantes des sites et habitats, et l’étude de la relation microchimie otolithe/environnement.
Les otolithes entiers sont nettoyés en conditions ultra-propres suivant la procédure de
Warner et al (2005) afin d’éliminer les contaminants de surface et les résidus de matière organique (Fig. 2.6). Lors de cette procédure, les otolithes sont individuellement déposés dans
les puits d’une plaque en plastique. Un premier bain d’une heure est effectué dans une
solution 50/50 d’eau oxygénée (qualité Suprapur® (H2O2, 30%)) et de soude (NaOH, 0,1
mol.L-1). Durant les 5 dernières minutes, la plaque est déposée dans une cuve à ultrasons. Les
otolithes sont ensuite rincés cinq fois cinq minutes dans des bains successifs d’eau ultra-pure.
Ils sont ensuite placés à sécher sous une hotte à flux laminaire (conditions HEPA 100), et
conservés jusqu’à analyses ICPMS dans des microtubes labellisés et préalablement
décontaminés à l’acide.
4.2. Coupes transversales
Des coupes transversales d’otolithes sont préparées de manière à découvrir l’ensemble de la vie des poissons, du noyau jusqu’au bord de l’otolithe, puis sont montées en lames minces.
La coupe transversale ne peut être réalisée directement et les otolithes doivent être inclus dans
des blocs de résine (Fig. 2.6). La résine utilisée est une résine Araldite® 2020, composée
d’une résine et d’un agent durcisseur mélangés en proportions égales (100g : 30g).
Préalablement aux inclusions des otolithes, une petite quantité de résine est répartie dans les
puits d’un moule en silicone. Ceci permet de réaliser les fonds qui sont laissés à sécher durant 12h au minimum. Ensuite, les otolithes sont déposés individuellement, face convexe sur le
fond, et recouverts de résine. L’ensemble des blocs est placé sous une hotte pendant quelques heures avant d’être placé dans une étuve à 50°C durant 24h.
32
Une fois les blocs de résine secs, des sections transversales fines contenant le noyau de
l’otolithe sont réalisées à l’aide d’une scie circulaire diamantée (Buehler®). La coupe fine
réalisée est collée avec de la résine adhésive Crystalbond 509® sur une cale en verre, ellemême fixée sur une lame porte-objet (Secor 1992). La section est polie à l'aide de disques de
polissage successifs aux grains de plus en plus fins (de 900 à 1200 grains/cm2). Le polissage
s’affine avec des films abrasifs (de 9, 3 puis 1µm) et deux solutions diamantées (1 et 0,3 μm) jusqu’à obtenir une section plane passant par le noyau de l'otolithe. Pour finir, les coupes sont nettoyées à l’eau ultra-pure, séchées sous la hotte à flux laminaire (HEPA 100) et conservées
dans des boites plastiques.
Figure 2.6 : Protocole de préparation des otolithes pour analyses in toto et analyses en
transects.
33
4.3. Echantillons de sédiment
Le protocole de préparation des échantillons de sédiment suit le protocole de Totland et
al. (1992). L’objectif est de récupérer la fraction fine (< 63µm) des sédiments car c’est au niveau de cette fraction granulométrique que la majorité des éléments sont biodisponibles.
Une fois décongelés, les échantillons sont tamisés sur une maille de 63µm à l’aide d’eau ultra-pure afin de récupérer uniquement la fraction fine du sédiment. La fraction fine et l’eau sont récupérées dans un bécher puis placées dans une étuve à 100°C jusqu’à complète
évaporation de l’eau. Une fois sèche, la fraction fine est broyée au mortier en agate, conservée dans un sachet plastique propre et placée dans un endroit sec, à l’abri de la lumière.
5. Analyses microchimiques
5.1. Principes de fonctionnement de l’ICP-MS
Les signatures microchimiques sont caractérisées par méthode de spectrométrie de masse
grâce à un spectromètre de masse à source plasma haute résolution ou ICP-MS (Inductively
Coupled Plasma Mass Spectrometer) (Fig. 2.7). Cet appareil permet de déterminer la
composition en éléments traces d’échantillons liquides en mode solution (SB- ICP-MS) ou
solides en mode laser (LA- ICP-MS). L’analyse des échantillons par ICP-MS se divise en
quatre étapes :
o Phase d’introduction et de nébulisation.
o Phase d’ionisation.
o Phase de séparation en masse et charge.
o Phase de détection.
34
Figure 2.7 : Photo d’un système LA-ICP-MS, Environmental Analytical Chemistry Unit,
Research Institute for the Environment and Livelihoods, Charles Darwin University.
La première phase dépend du type d’échantillons. Les échantillons liquides sont injectés dans le système par une aiguille de prélèvement automatique en téflon et entraînés à l'aide
d'une pompe péristaltique jusqu'au nébuliseur qui permet de faire passer l'échantillon à l'état
d'aérosol liquide (microgouttelettes de quelques μm). Pour les échantillons solides, l’appareil est couplé à un système d’ablation laser qui dégage les éléments traces en provoquant une ablation à la surface des otolithes. Ces éléments traces sous forme d’aérosols, sont véhiculés grâce à un courant d’argon et d’hélium. A partir de la phase d’ionisation, le principe de l’appareil est le même pour les deux types d’échantillons. Les aérosols passent au niveau d’une torche à plasma d’argon et sont vaporisés, dissociés, atomisés et ionisés sous l’effet d’une température élevée (entre 5000 à 8000°C). Pour la phase de séparation, une lentille ionique conduits les ions vers le spectromètre de masse quadripolaire où ils sont séparés selon
leur charge (z) et leur masse (m). Seuls les ions sélectionnés au préalable sont transmis au
détecteur dans lequel ils percutent un multiplicateur d’électrons à dynodes. Pour chaque ion qui percute la paroi des dynodes, ces dernières émettent des électrons. La série de dynodes de
l’appareil va provoquer un effet « boule de neige » et les électrons émis heurtent au final un
collecteur équipé d’un préamplificateur. C’est la phase de détection où les électrons émis sont dénombrés par le collecteur. Le signal se traduit en nombre de coups par seconde (nombre
d'impulsions) et est directement proportionnel à la concentration de l’élément étudié dans l’échantillon. Pour finir, une interface informatique assure le transfert des données afin qu'elles soient traitées.
35
Les résultats en coups par seconde doivent être convertis en concentrations, exprimées en
ppm (i.e. partie par millions) correspondant à un rapport de 10-6 (µg.g-1). Les concentrations
sont calculées à l’aide d’un standard interne dont la concentration dans l’échantillon est connue. Pour les échantillons solides carbonatés comme les otolithes, le standard interne
utilisé est le calcium alors que pour les échantillons liquides, le standard interne est une
solution d’Indium/Bismuth. De plus, pour les échantillons solides, un standard externe (NIST
612) est utilisé au début et à la fin de chaque procédure pour obtenir des courbes de
calibration et palier la dérive temporelle de l’appareil. Pour les sédiments en solution, deux standards externes sont analysés en fin de procédure (UBR et BEN).
5.2. Choix des éléments analysés
Outre le Sr, le Ba et le Mn, des éléments tels que le Fe, Pb, Li, Mg, Cu et Ni sont
susceptibles de servir de marqueurs caractéristiques de l’environnement (Campana 2005).
Afin d’augmenter l’efficacité discriminatoire entre les sites, le maximum d’éléments possible est retenu pour les analyses d’où la notion d’« empreinte multi-élémentaire ». L’ICP-MS va
permettre de déterminer la composition en éléments suivants : Li, B, Rb, Sr, Mo, Cd, Sn, Ba,
Pb, Th, U, Mg, Ca, Ti, V, Cr, Mn, Ni, Cu, Zn via la mesure des isotopes suivants : 7Li,
11
B,
85
Rb, 88Sr, 95Mo, 111Cd, 117Sn, 138Ba, 208Pb, 232Th, 238U, 25Mg, 43Ca, 44Ca, 47Ti, 51V, 52Cr, 55Mn,
60
Ni, 65Cu, 66Zn.
5.3. Analyses chimiques des otolithes
5.3.1. Analyses laser sur otolithes in toto
Contrairement à la technique de dissolution des otolithes expérimentée dans plusieurs
études de comparaisons interspécifiques (Gillanders et al. 2003, Swearer et al. 2003), la
technique de l’ICP-MS couplé à un laser (LA-ICP-MS) permet de s’affranchir d’une différence d’incorporation des éléments due à une fréquentation d’environnements distincts
au cours de la vie des espèces étudiées. La technique du laser permet de faire une ablation du
bord de l’otolithe et donc une analyse des éléments les plus récemment incorporés au sein de
l’otolithe alors que la dissolution analyse l’ensemble de la vie. De ce fait, la microchimie du
bord des otolithes devrait refléter plus précisément l’exposition des poissons aux conditions
environnementales du lieu de capture des individus et de la collecte des échantillons d’eau et de sédiment. Les analyses sont toujours réalisées au même site sur l’otolithe entier c'est-à-dire
36
au niveau du post-rostre (Fig. 2.8). Cette précaution est importante afin de s’assurer que les différences observées ne soient pas dues à des variations dans le site d’analyse (Hamer &
Jenkins 2007). Une face d’une lame porte-objet est recouverte de scotch transparent double
face sur lequel les otolithes sont collés. Ils sont agencés de manière à ce que les post-rostres
soient alignés horizontalement. Cette démarche permet de simplifier la recherche du site
d’analyse une fois la lame installée dans la chambre du laser. L’analyse par laser ICP-MS se
constitue d’un tir laser sur le bord postérieur des otolithes (Fig. 2.8). La procédure d’analyse consiste en 30 secondes de blanc (laser non actionné) suivie de 2 minutes d’acquisition avec un faisceau laser de 90 μm de diamètre, 5 Hz de fréquence et 15 joules.cm-2 de puissance.
En moyenne, 80 ablations laser peuvent être réalisées par journée d’ICP-MS, soit 80
otolithes (ou individus) analysés par jour.
Figure 2.8 : Illustration de la méthode d’analyse in toto. (A) montage des otolithes entiers sur
lame et (B et C) position de l’analyse laser sur otolithe in toto.
5.3.2. Analyses laser en transects
Cette méthode permet de reconstruire l’histoire de vie environnementale des poissons. Les otolithes destinés aux analyses laser par transects sont préparés en section transversale
fine comme détaillée précédemment. Les transects se présentent sous la forme d’une succession d’ablation laser d’un diamètre de 90µm tout les 120 µm, du noyau vers le bord de
l’otolithe le long de l’axe principal de croissance (Fig. 2.9). La procédure d’analyse consiste en 30 secondes de blanc (laser non actionné) suivie de 2 minutes d’acquisition avec un faisceau laser de 90 μm de diamètre, 5 Hz de fréquence et 15 joules.cm-2 de puissance. Le
nombre de transects réalisé en une journée d’analyse dépend de la taille de l’otolithe (i.e. distance noyau-bord). En moyenne, 3 transects constitués de 25 ablations chacun peuvent
êtres effectués par jour (75 ablations par jour).
37
Figure 2.9 : Illustration de la méthode d’analyse en transects. (A) Coupes transversales positionnées dans la chambre du LA-ICP-MS et (B) transect sur une coupe transversale
d’otolithe colorée au bleu de Toluidine pour ageage.
5.4. Analyses chimiques de l’environnement
Les échantillons d’eau et de sédiment sont analysés par ICP-MS en mode solution (SBICP-MS) et les éléments traces dosés sont les mêmes que pour les otolithes.
Avant analyse, les échantillons d’eau sont dilués par 40 afin d’obtenir une solution de salinité proche de 1‰ dans un milieu matriciel de 2,5% de HNO3 contenant 1,0% d’Indium, standard interne qui permet de corriger la dérive de l’ICP-MS. Pour cela, 0,25 ml de
l’échantillon d’eau de mer est dilué dans 9,4 ml d’H2O ultrapure, 0,25 ml d’HNO3 et 1 ml
d’Indium.
Le passage à l’ICP-MS des sédiments nécessite une mise en solution. Plusieurs attaques
acides successives sont réalisées dans des béchers en Téflon® sur 100mg de portion fine pour
chaque échantillon (Fig. 2.10). Les acides utilisés sont de grade Suprapur®, l’eau utilisée est ultrapure et la totalité du protocole de Totland et al. (1992) se réalise sous hotte aspirante. La
procédure consiste à deux phases de digestion acide forte à l’aide d’HNO3 (69%) et d’HF (40%). Ensuite, deux phases de digestion acide modérées à l’HNO3 (69%) uniquement. Une
reprise est effectuée et diluée avec de l’H2O. Un aliquote d’1/10 ml est réalisé. La solution
finale de sédiment est diluée par 1000. Pour les analyses, la solution introduite dans l’appareil est constituée de 0,25 ml de la solution finale diluée avec 9,4 ml d’H2O, 0,25 ml d’ HNO3 et 1
ml d’Indium. 38
Figure 2.10 : Protocole de mise en solution des sédiments par acidification.
6. Traitement des données
6.1. Transformation des données microchimiques
Les données brutes issues de l’analyse ICP-MS sont exprimées en coups par seconde
(cps) et doivent être transformées en concentrations exprimées en parties par millions (ppm).
La méthode utilisée est celle de Longerich et al. (1996) et a été réalisée à l’aide d’un script élaboré sous le logiciel statistique R (R Development Core Team 2011). Pour chaque
élément :
39
o Les données aberrantes (outliers ou données extrêmes), définies comme toutes valeurs
supérieures à trois fois l’écart interquartile, ont été exclues (Tukey 1977). C’est une procédure classique en microchimie car les ICP-MS sont des appareils très sensibles
qui produisent régulièrement des valeurs ponctuelles aberrantes appelées « spikes ».
o Les concentrations et les limites de détection (LOD) ont été calculées suivant la
méthode de Longerich et al. (1996).
o Les éléments conservés pour les analyses statistiques ont été sélectionnés suivant deux
critères : 1) les concentrations élémentaires dans les otolithes doivent être supérieure à
la LOD dans 70% des analyses dans au moins un des habitats (ex. la mangrove) ou un
des sites (ex. Ouano). 2) le coefficient de variation des concentrations mesurées au
sein des standards externes (NIST 612) doit être inférieur à 10% pour chaque élément
(Chittaro et al. 2004, Chittaro et al. 2006). Les quelques valeurs inférieures à la LOD
et conservées dans les analyses ont été fixées à zéro.
o Dans le but de réduire les possibles variations causées par l’utilisation de deux ICPMS différents, les concentrations élémentaires ont été standardisées par ligne (i.e. par
échantillon) afin d’obtenir des compositions élémentaires relatives (en pourcentages
de chaque élément dans l’échantillon).
6.2. Analyses statistiques
Les analyses statistiques utilisées sont adaptées à des données ne suivant pas les
hypothèses de normalité et d’homoscédasticité, ainsi qu’à des plans d’échantillonnage non équilibrés. Toutes les analyses statistiques ont été réalisées à l’aide du logiciel R (R
Development Core Team 2011).
6.2.1. Analyses descriptives
Des analyses de variances (ANOVA) à un et deux facteurs sont utilisées afin de tester les
différences spatiales de compositions élémentaires (analyses multivariées MANOVA) et de
concentrations élémentaires (analyses uni-variées ANOVA). Les ANOVA classiques exigent
de respecter des conditions d’utilisation telles que la normalité des résidus et l’homogénéité des variances. Si les données ne répondent pas à ces postulats de base, elles sont en général
transformées afin de ramener leur distribution à la normalité et d’homogénéiser la variation. A l’inverse, les méthodes d’ANOVA par permutations (PERMANOVA) permettent de se libérer de ces conditions d’utilisations. Elles construisent leur propre loi de distribution en utilisant
40
des permutations aléatoires du jeu de données (Legendre & Anderson 1999, Anderson 2001).
La technique des PERMANOVA a été utilisée dans ce travail.
6.2.2. Discrimination spatiale
La méthode des Random Forest (forêts aléatoires en français) a été utilisée afin de
discriminer les échantillons à différentes échelles spatiales à partir des compositions
élémentaires des otolithes et de l’environnement. Il existe différentes méthodes de
classification/prédiction utilisées dans les études de microchimie des otolithes : les Réseaux
de Neurones (ANN) et les Analyses Discriminantes Linéaires (LDA) et Quadratiques (QDA).
Dans une étude comparative, Mercier et al. (2011) ont cependant démontré que les Random
Forest (RF) constituent la méthode statistique la plus adaptée aux données de microchimie des
otolithes. Elle présente l’avantage de se libérer des conditions de normalité et
d’homoscédasticité requises avec les autres méthodes statistiques. La pertinence de l’utilisation des RF a ensuite était démontrée au cours d’études de connectivité utilisant des jeux de données de microchimie des otolithes provenant des lagunes de la région du Golfe du
Lion (Mercier et al. 2012, Tournois et al. 2013).
La méthode des Random Forest (RF) est basée sur la construction d’arbres de décision multiples. Chaque arbre de décision est construit à partir du tirage aléatoire d’un sous ensemble du jeu de données initial (classiquement les deux tiers), appelé « jeu de
calibration ». Un arbre est constitué de nœuds correspondant à un partitionnement des
observations en fonction de la valeur des variables. Pour les RF, la procédure de
partitionnement aux nœuds est partiellement aléatoire. À chaque nœud, il y a un sous échantillonnage aléatoire des variables utilisées (ici les éléments chimiques) pour scinder les
observations (ici les habitats ou les sites). Le tiers du jeu de données initial qui n’est pas utilisé pour la construction de l’arbre de décision est appelé « jeu test » car il est utilisé pour
estimer la capacité de prédiction de cet arbre indépendamment du jeu de calibration utilisé
pour le construire. Ainsi, la classe (i.e. habitat ou site) de chaque observation du jeu test est
connue (i.e. lieu d’échantillonnage) et comparée à la classe prédite par l’arbre de classification compte tenu de la valeur des variables pour chaque observation. Ceci permet d’obtenir une valeur indépendante du taux d’erreur de l’arbre de classification des observations. Il y a donc
deux procédures de tirage aléatoire dans les RF : chaque arbre est construit à partir d’un jeu aléatoire de calibration, et chaque nœud est calculé à partir d’un tirage aléatoire des variables
41
de partitionnement. Cette procédure est réalisée de manière itérative afin de construire une
forêt d’arbres, le nombre d’arbres réalisés étant défini par l’utilisateur (dans ce travail, 5000
arbres sont construits à chaque RF). Cette construction d’arbres multiples permet de prédire la
classe de chaque observation du jeu de données initial à plusieurs reprises, chaque arbre
constituant un vote pour une classe donnée. La prédiction finale d’une observation correspond
à la classe qui reçoit la majorité des votes sur le total des arbres de la forêt. De manière
importante, le pourcentage de vote de cette prédiction est connu et constitue la probabilité
associée à cette prédiction finale. Ainsi, les RF non seulement classifient les observations en
fonction de la valeur des variables mais associent une probabilité (% vote) à chaque
prédiction.
6.2.3. Echelles spatiales et organisationnelles
La méthode des RF a été utilisée afin de discriminer les échantillons à trois échelles
spatiales à partir des compositions élémentaires des otolithes et de l’environnement. Dans un premier temps, les discriminations ont été réalisées à l’échelle globale. Cette échelle est
constituée des données provenant des 11 sites autour de la Nouvelle-Calédonie et les
différences ont été testées entre :
o Habitats (mangrove versus récif barrière interne).
o Sites pour un habitat donné (mangroves, récifs barrières internes).
Ensuite, les discriminations ont été réalisées à l’échelle régionale. Cette échelle
comprend les données issues des 4 sites ateliers de la côte ouest et les différences ont été
testées entre:
o Habitats (mangrove vs récifs frangeant, intermédiaire et barrière interne).
o Sites pour un habitat donné (mangroves, récifs frangeants, récifs intermédiaires, récifs
barrières internes).
Enfin, les discriminations ont été réalisées à l’échelle locale définie comme l’échelle spatiale la plus fine. Cette échelle correspond à chaque site atelier et les différences ont été
testées:
o Entre habitats (mangrove vs récifs frangeants, intermédiaires et barrière interne).
42
Ces analyses à différentes échelles spatiales ont été réalisées pour les niveaux
organisationnels suivants :
o L’espèce.
o Multi-spécifique (toute espèce).
o Groupe taxonomique (famille).
o Groupe fonctionnel (régime alimentaire).
Lors de la construction d’arbres de classification, certains éléments chimiques sont informatifs car ils apportent de l’information utile à la classification, et d’autres éléments n’apportent que du bruit. L’élimination de ces éléments non-informatifs permet d’augmenter la précision des classifications tout en réduisant le nombre d’éléments constituants les
signatures chimiques (Mercier et al. 2011) . A chaque échelle spatiale et organisationnelle
testée, les RF ont donc été réalisées avec toutes les combinaisons d’éléments possibles. La meilleure combinaison d’éléments retenue est celle qui correspond au meilleur taux de
classification correcte avec le plus petit nombre d’éléments.
43
44
Chapitre 3 : Quantification du pouvoir discriminant des
signatures multi-élémentaires mono-spécifiques et multispécifiques dans le lagon de Nouvelle-Calédonie
45
46
Chapitre 3 : Pouvoir discriminant des signatures
Chapitre 3 : Quantification du pouvoir discriminant des
signatures
multi-élémentaires
mono-spécifiques
et
multi-
spécifiques dans le lagon de Nouvelle-Calédonie
Dans un milieu oligotrophe tels que les récifs coralliens, la discrimination spatiale des
habitats par microchimie des otolithes est réputée limitée par l’absence de contraste
environnemental et donc par une chimie de l’environnement relativement homogène (Patterson et al. 2004, Berumen et al. 2010).
En Nouvelle-Calédonie, la microchimie des otolithes a été étudiée à deux reprises et les
résultats issus de ces premières études se sont avérés très prometteurs (Labonne et al. 2008,
Sigura 2009). Il semblerait notamment que le potentiel informatif et discriminant de l’outil
microchimique soit élevé sur le territoire. L’étendue géographique de la Nouvelle-Calédonie,
sa spécificité géologique, la superficie de son lagon et l’hétérogénéité des habitats qu’il abrite seraient donc susceptibles d’induire des gradients environnementaux marqués. Ces gradients pourraient ensuite se traduire dans les otolithes de poisson par des signatures microchimiques
caractéristiques des environnements traversés. Si tel est le cas, ces signatures pourraient être
utilisées pour étudier la connectivité entre les différents environnements et habitats du lagon
calédonien. Néanmoins, avant leur utilisation, il est nécessaire de démontrer l’existence et de caractériser le pouvoir discriminant de telles signatures en Nouvelle-Calédonie.
Ce chapitre s’insère dans l’axe méthodologique de la thèse et a pour objectif de
caractériser les signatures microchimiques dans les otolithes de poisson en Nouvelle
Calédonie. Il s’agira notamment de :
1) Caractériser les variations spatiales de la microchimie des otolithes à l’échelle des sites et habitats de la Nouvelle-Calédonie.
2) Evaluer la précision des signatures microchimiques mono-spécifiques pour différentes
espèces présentant des traits de vie contrastés.
3) Etudier la possible utilisation des signatures microchimiques multi-spécifiques comme
proxy des signatures mono-spécifiques.
47
Ce travail de caractérisation de la microchimie des otolithes est primordial pour la suite
de la thèse, et notamment pour le développement de l’axe écologique. Il fournit la base de la compréhension du pouvoir microchimique existant en Nouvelle-Calédonie et sera complété
par le chapitre méthodologique suivant, les liens entre la microchimie de l’environnement et des otolithes.
48
Manuscrit A: Assessing the power of otoliths multi-elemental signatures in
the coral reefs and mangroves of New Caledonia
C. Paillon, 1, 2, L. Wantiez 2, 3, M. Labonne 4, L. Vigliola 1
1
Institut de Recherche pour le Développement (IRD), UR227 CoRéUs, Laboratoire Excellence LABEX corail,
Noumea, New Caledonia
2
Université de Nouvelle-Calédonie (UNC), LIVE (EA4243), Noumea, New Caledonia
3
Aquarium des Lagons, Noumea, New Caledonia
4
Institut de Recherche pour le Développement (IRD), LEMAR (UMR 6539), Institut Universitaire Européen de
la Mer, Plouzané, France
Corresponding author: email: christelle.paillon@ird.fr
Abstract
Otoliths microchemistry is the only method available to reconstruct lifetime movements of
fish and lifetime connectivity among habitats at individual level. Prior to use microchemistry
as a tool to study connectivity, elemental fingerprints need to be characterized and their
accuracy evaluated. The objectives of the present study were (1) to characterize spatial
variations of elemental fingerprints in New Caledonia, (2) to assess the spatial scale at which
correct site and habitat classifiers could be built using elemental fingerprints of numerous
species, and (3) to evaluate the accuracy of multi-specific elemental classifiers as a proxy of
mono-specific classifiers according to taxonomic or ecological similarity among species.
Using Random Forest classification method, we found high correct classification rates at
habitat level, specifically between mangrove (86.4%) and barrier reef (98.8%) for Lutjanus
fulviflamma. Correct classification rates at the station level were lower than at the habitat
level, except among mangrove stations (between 63 and 100% of correct classification).
49
Discriminations were enhanced by the chemical heterogeneity of coastal habitats and the
chemical contrast between distant habitats. Use of multi-specific classifiers as substitutes of
mono-specific classifiers was generally not relevant, even when multi-specific signatures
were built with ecological or functional related species. However, when habitats were much
contrasted, as it is the case for mangroves and barrier reefs, then multi-specific signatures
appeared very relevant. Our results imply that studies estimating fish connectivity using
otolith microchemistry require a sampling of all species concerned unless chemical contrast
among habitats or stations is very high.
Key words: multi-element, LA-ICP-MS, otoliths, Random Forest, multi-specific
50
1. Introduction
Tropical seascapes harbor a high diversity of fish assemblages spread over multiple
ecosystems such as mangroves, seagrasses and coral reefs. Actual increasing anthropogenic
perturbations weaken these coastal ecosystems by threatening existing connectivity between
local populations of fish within tropical seascapes. Therefore, identifying and quantifying
connectivity in tropical seascapes has become a priority of conservation management
(Almany et al. 2009, Sale et al. 2010). However, connectivity between fish populations
remains a not fully understood ecological process, principally due to a challenging acquisition
of empirical data of fish lifetime movements. Among methods for determining fish
movements, otoliths microchemistry is the most promising to quantify connectivity by
reconstructing lifetime movements (Fairclough et al. 2011, Mercier et al. 2012), identifying
natal origins (Swearer et al. 2003, Standish et al. 2008), putative nurseries (Brown 2006b,
Mateo et al. 2010) and habitat utilization (Forrester & Swearer 2002) of fish stocks.
Otoliths are paired calcareous structures located in the inner ear of all bony fishes, implicated
in equilibrium and sound reception. They grow continuously by daily accretions of calcium
carbonate (CaCO3) originating from the surrounding waters. These deposits permit the
incorporation of trace elements also originating from the surrounding waters. Otoliths being
metabolically inert, the trace elements incorporated are never reworked or resorbed (Campana
1999). These key-properties of otoliths qualify them as natural recorders of water masses
crossed by the fish during his lifetime. Microchemistry method consists in measuring the
elemental composition of otoliths which, if it reflects the physicochemical properties of the
surrounding water at a given time and area, will define an elemental fingerprint. Then, the
chemical fingerprint can be used as a natural tag of a geographic location (Bath et al. 2000,
Elsdon & Gillanders 2003a, Sturrock et al. 2012).
Prior to use otoliths microchemistry as a quantifier of actual connectivity, there is a need to
define elemental signatures of species for a maximum of sites and to obtain a high level of
discrimination between them (Hamer & Jenkins 2007). For a single species, spatial variations
of elemental signatures could either allow good discrimination among study sites (Gillanders
& Kingsford 2000, Swearer et al. 2003) or could not revealed significant differences
(Patterson et al. 2004, Berumen et al. 2010). Benefit or loss of discrimination between
locations seems to be related to the number of trace elements measured. Besides main trace
51
elements like Sr, Ba, Mg and Mn classically used in otolith microchemistry studies, Campana
(2005) suggested that others elements such as Fe, Pb, Li, Cu, Ni, Al, Co, Zn, Ag, Cd, and Sn
could also be effective markers of environment.
Unlike other coastal habitats with high terrigeneous inputs (Gillanders & Kingsford 2000,
Forrester & Swearer 2002, Brown 2006a, Cuveliers et al. 2010, Tournois et al. 2013), coral
reefs are oligotroph systems with a limited variation of environmental chemistry of water
masses (Patterson et al. 2004, Chittaro et al. 2006). Consequently, studying spatial variability
of chemical fingerprints based exclusively on a limited number of trace elements is generally
insufficient to discriminate different sites (Berumen et al. 2010). For example, Patterson and
Kingsford (2004) studied spatial variability using Sr, Ba and Mn on the Australian Great
barrier reef. Discrimination was possible at a large spatial scale (several hundred of km) but
not at a smaller scale (ten km). They concluded that the coral reefs of the Great Barrier Reef
could not be considered as different microchemistry units because of the low terrigeneous
influence leading to a lack of variation of environmental parameters. On the opposite,
studying multi-elemental signatures provided a higher degree of discrimination (Chittaro et al.
2004, Labonne et al. 2008).
In this respect, New Caledonia lagoon is of special interest because it constitutes a tropical
seascape including coral reefs close to coast and under high terrigeneous influence,
originating from numerous freshwater and anthropogenic inputs (Baille et al. 2012).
Moreover, New Caledonia presents a mining history with past and actual exploitations of Ni,
Co and Cr (Paris 1981). This specificity may generate heterogeneity in the environmental
chemistry, related to the distance offshore, the habitat and the site (Labonne et al. 2008,
Andréfouët et al. 2009).
Mono-specific elemental fingerprint can be difficult to obtain because of logistic constraints
or conservation issues depending on the studied species. In this case, a multi-specific
elemental fingerprint will be easier to obtain. However, heterogeneity of trace elements
incorporation among species is not a well-understood process. Consequently, the use of multispecific signature as a proxy should not be considered at random and should be carefully
evaluated. Study of inter-specific variation of elemental signature is a way to adjust the
accuracy of multi-specific signatures when selecting species by taxonomic or functional
52
similarity. Several studies focused on inter-specific variation of otoliths fingerprints
(Gillanders & Kingsford 2003, Chittaro et al. 2006, Hamer & Jenkins 2007). Comparisons
between elemental signatures of different species generally highlighted significant
differences. However, similar patterns in spatial variations between species could be observed
and revealed that elemental signatures were more similar between taxonomically closed
species than distant ones (Swearer et al. 2003). Nevertheless, these taxonomically closed
species have to share ecological traits such as diet and/or habitat.
The aims of this study were: (1) to describe the spatial variations of elemental compositions of
otoliths in New Caledonia and the relative contribution of elements; (2) to assess the spatial
scale at which elemental fingerprints can build correct classifiers for several species; (3) to
evaluate the accuracy of multi-specific elemental classifiers as a proxy of mono-specific
classifiers according to taxonomic or ecological similarity.
2. Material and methods
2.1. Sampling protocol
New Caledonia is a French archipelago located in the South West Pacific, 1500 km east of
Australia (Fig. 3.1). The archipelago is composed of a main island and several groups of
smaller islands. The main island is around 17 000 km2, 400 km long and 60 km wide at its
maximum. The main island is surrounded by the largest lagoon of the world (31 336 km ),
delimited by a barrier reef of 1744 km long, the second longest in the world after the GBR
(Andréfouët et al. 2009). The distance between the coast and the barrier reef varies from 1 to
65 km (Paris 1981) and some parts of the lagoon are registered on the Unesco heritage list
since 2008. New Caledonia lagoon gathers several types of ecosystems constituting a tropical
seascape such as mangroves, seagrasses, algal beds, bare soft bottoms and coral reefs
(fringing, intermediate and barrier).
53
Figure 3.1: (A) New Caledonia geographic situation in the South West Pacific. (B) Study
sites: circles and triangles indicate sites with 2 and 4 habitats sampled, respectively detailed in
C and D (with M: mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier).
Fish were sampled for otolith analysis in different mangrove and reef habitats at 11 sites
around the main land of New Caledonia in three years (2009, 2010 and 2011) (Fig. 3.1 and
Tab. 3.1). In reef habitats, large species were collected by spear fishing and small species
were anesthetized using clove oil and collected with handnets. In mangroves, gillnets were
deployed at high tide along the mangrove maritime front and fish were collected with
handnets in a few cm of water and/or on the ground at low tide. Different mesh sizes were
used in order to catch species of different length and at different ontogenetic stage.
Immediately after collection, fishes were stocked in a freezer (at -20°C) until dissection.
The first sampling took place during austral summer 2009 in the mangroves and inner barrier
reefs of two sites (S2 and S5). The second sampling has been carried out during austral winter
2010 in the mangroves and inner barrier reefs of eleven sites around the main island. Fringing
54
and intermediate reefs were also sampled at four sites (S2, 3, 5 and 11). These four sites were
designated as “west coast sites” and replicated during the third sampling in austral winter 2011. During the first and second samplings, a multi-specific collect has been carried out with
respectively 44 and 25 species sampled (Tab. 3.1). The third sampling focused on four target
species: Dascyllus aruanus, Ctenochaetus striatus, Lutjanus fulviflamma and Siganus
lineatus.
Table 3.1: Sample size and otolith analyses; (PSO: Pôle de Spectrométrie Océan; RIEL:
Resources Institute of Environmental Livelihoods).
Year
Site
Habitat
n stations
n species
LA-ICP-MS
n otoliths
2009
2
2 (M, BR)
4
44
PSO
209
2010
7
4
2 (M, BR)
4 (M, FR, IR, BR)
14
16
25
PSO
RIEL
114
595
2011
4
4 (M, FR, IR, BR)
16
4
PSO
RIEL
130
60
Multi-specific classifiers build thereafter included otolith elemental compositions of all
species caught during the three samplings. Mono-specific classifiers were realized for the
species showing the highest redundancy among the three samplings. They were built for the
eight following species: Dascyllus aruanus, Chaetodon lunulatus, Ctenochaetus striatus,
Gerres oyena, Gnathodentex aureolineatus, Lutjanus fulviflamma, Scolopsis bilineata and
Siganus lineatus. Taxonomic and functional details of the eight species are summarized in
table 3.2.
Table 3.2: Summary of species characteristic used to build mono-specific signatures (C:
carnivore, MC: micro carnivore, H: herbivore, Z: zooplankton feeders).
Specie
Family
Home range
Diet
Lutjanus fulviflamma
Lutjanidae
Mobile
C
Ctenochaetus striatus
Acanthuridae
Sedentary
H
Dascyllus aruanus
Pomacentridae
Sedentary
Z
Siganus lineatus
Siganidae
Very mobile
H
Chaetodon lunulatus
Chaetodontidae
Sedentary
MC
Gerres oyena
Gerreidae
Mobile
C
Gnathodentex aureolineatus
Lethrinidae
Sedentary
C
Scolopsis bilineata
Nemipteridae
Mobile
C
55
2.2. Otolith preparation and chemical analyses
All material used for otolith handling was previously decontaminated in 5% ultrapure nitric
acid bath during 24h, rinsed three times with ultrapure water (18.2 MΩ), dried and stored in
clean plastic bag under a laminar flow hood (HEPA 100).
Before extraction of otoliths, fish samples were unfrozen and each individual was measured
(Fork length, FL) to the nearest mm and weighted to the nearest g. Paired sagittal otoliths
were extracted using ceramic tweezers and ultrapure water.
The chemical signatures of habitats were characterized by analyzing the chemical
composition at the surface of otoliths. For mono-specific signatures, up to five otoliths were
randomly selected in each sample of the eight following species: Dascyllus aruanus,
Chaetodon lunulatus, Ctenochaetus striatus, Gerres oyena, Gnathodentex aureolineatus,
Lutjanus fulviflamma, Scolopsis bilineata and Siganus lineatus. For multi-specific signatures,
up to three otoliths were randomly selected for each species caught in each sample.
Otoliths were cleaned of adhering tissues following Warner et al (2005) method with a bath of
50/50 H2O2 (30%, Suprapur®) and NaOH (0.1 mol.L-1, Suprapur®) during 1 hour. Otoliths
were sonicated during the last 5 min of the bath, and then rinsed 5 times with milli-Q water
for 5 min, dried under a laminar flow hood (HEPA 100) and stored in individual plastic vials.
Otoliths were chemically analyzed by Laser Ablation Inductively Coupled Plasma Mass
Spectrometry (LA-ICP-MS) at the Pôle de Spectrométrie Océan (Institut Universitaire
Européen de la Mer, Brest, France) using a Thermo Element 2 coupled to a laser 193 nm
CopexPro 102 Coherent, and at the Resources Institute of Environmental Livelihoods
(Charles Darwin University, Darwin, Australia) using Agilent 7500ce coupled to a UP – 213
nm laser ablation system (Tab. 1). Both ICP-MS were operated at low resolution using argon
as the carrier gas. Both laser systems parameters were set on a 90 µm laser beam diameter and
a frequency of 5 Hz. Each analysis lasted 120 s including 30 s of background and 90 s of
ablation (laser activated). To maximize discrimination between habitats, 20 isotopes were
measured: 7Li,
51
52
55
11
B,
60
85
Rb, 88Sr, 95Mo,
65
66
V, Cr, Mn, Ni, Cu and Zn.
56
111
Cd,
117
Sn,
138
Ba,
208
Pb,
232
Th,
238
U,
25
Mg,
43
Ca,
47
Ti,
To reveal the latest elements incorporated from the habitat where the fish was caught, one
laser ablation was done on the otolith surface. To standardize the analyses, the ablation was
always done at the same location at the tip of the post rostrum on the otolith posterior side.
Calcium was used as an internal standard to compensate possible variation due to differences
in quantity of material ablated. To correct for instrument drift, an external standard (National
Institute of Standards and Technology, NIST, 612) was analyzed twice at the beginning and at
the end of each session and also every ten ablations.
Raw ICPMS data (counts per second, CPS) were cleaned of outliers, defined as values higher
than three times the inter-quartile distance during both the blank and the ablation windows
(Longerich et al. 1996, Heinrich et al. 2003). Cleaned data were then transformed in
elemental concentrations (parts per million, ppm) and limits of detection (LOD) were
calculated following Longerich et al. (1996). Any elemental concentrations lower than the
LOD was set to zero. Only the elements that met the following two criteria were selected in
statistical analyses: 1) elemental concentrations in otoliths had to be higher than the LOD in
70% of the otoliths in at least one habitat or one site and 2) the coefficient of variation of
elemental concentrations in the NIST 612 had to be less than 10% (Chittaro et al. 2004,
Chittaro et al. 2006). The use of two different ICP-MS and the time intervals between two
sessions realized with the same ICP-MS may influence to some degree the absolute values of
concentration. To address this potential issue, elemental concentrations were transformed in
percent of all measured elements.
2.3. Statistical analyses
All statistical analyses were performed using R statistical software (R Development Core
Team 2011). To describe spatial variation in multi-specific and mono-specific elemental
composition of otoliths, non-parametric analysis of variance (PERMANOVA) and principal
component analysis (PCA) were performed. Multivariate two-way PERMANOVA (with 999
permutations) based on Euclidian distance were performed to test the effect of site and habitat
on the multi-elemental composition of otoliths (Legendre & Anderson 1999, Anderson 2001).
Non parametric PERMANOVAs were used since the assumptions of normality and
homogeneity of variance required by parametric MANOVAs were not met. The PCA was
used to visualize which element characterized the spatial variations of the elemental
57
composition of otoliths. Elemental compositions were arcsin√x transformed in the PCA to
equalize the variance and normalize the data.
Random forests (RF) algorithm (Breiman 2001) was performed to build habitat classifiers
based on elemental concentrations of otolith. RF classification method allows freedom from
normality and homoscedasticity assumptions (Breiman 2001) and is one of the most powerful
methods for classification of otolith chemical signature . With the RF classification method,
two-third of the data set is used to build a classification tree and the remaining one-third is
classified along this tree. The RF algorithm builds 5000 trees to ensure that every individual
gets predicted several times which permits to estimate classification accuracy (percent of
correct classification). RF was performed for all possible element combinations in order to
find the combination that showed the highest percent of correct classification. When several
best combinations were found, the one with the smaller number of elements was retained.
RF was performed at different spatial scales and for several organizational levels. Three
spatial scales were considered. The local scale focused on discriminations between habitats
(mangrove, fringing, intermediate and barrier reef) within each site of the west coast. The
regional scale focused on the west coast sites gathered and discriminations were done between
habitats (mangrove, fringing, intermediate and barrier reef) and between stations of a specific
habitat. The global scale focused on the 11 sites of the main island and discriminations were
done at the level of habitat (mangrove versus barrier reef) and between stations of a specific
habitat. Elemental signatures were explored at the species level for eight species (monospecific signatures) and at the multi-species level with signatures calculated from data on 53
species, by grouping species from the same family and by grouping species with the same
diet.
At all spatial scales, RF classifier were calculated for all 8 species for which we add enough
information and their accuracy estimated from the percentage of individuals correctly
classified into their actual location of collection. Then, classifiers were built with multispecific elemental compositions using all species, species grouped by family and species
grouped by diet. These multi-specific classifiers were used to predict the location of collection
of the 8 species for which mono-specific signatures were calculated. The multi-specific
predictions were then compared with the actual location of collection. That way, we could
58
indeed compare the accuracy of mono-specific and multi-specific classifier in predicting the
location of collection of fish from 8 species. When a multi-specific classifier was build for
comparison with a mono-specific classifier, the species under study was excluded from the
multi-specific dataset in order to avoid circularity. For instance, L. fulviflamma elemental
compositions were predicted using multi-specific signatures calculated from all species, all
Lutjanidae or all carnivores except L. fulviflamma, and these multi-specific predictions were
compared with predictions from mono-specific signatures calculated from L. fulviflamma
dataset only.
Pearson correlation test was performed to evaluate the correlation between classification
accuracies obtained with mono-specific classifiers and accuracies obtained with multi-specific
classifiers.
3. Results
3.1. Spatial characterization of otolith microchemistry
Twelve elements were retained in the statistical analyses: B, Ba, Cr, Mg, Mn, Pb, Rb, Sn, Sr,
Th, U and Zn. Only spatial characterization of otolith chemistry obtained with the multispecies dataset and one mono-species dataset (L. fulviflamma) are detailed below. Results for
the remaining seven mono-species datasets are summerised below and detailed in Appendix 1
and 2.
Using multi-species otolith datasets of west sites, two-way multivariate PERMANOVA
showed a significant interaction of factors (site x habitat, p-value = 0.001), indicating that
differences among habitats were not consistent among sites (Tab. 3.3). Indeed, the PCA
highlighted strong overlap among reef habitats (Fig. 3.2A) and sites (Fig. 3.2B). Similar
results were obtained at species level with L. fulviflamma otoliths datasets of west coast sites.
Two-way multivariate PERMANOVA indicated a significant interaction between factors (site
x habitat, p-value = 0.013) and a PCA highlighted strong overlap between habitats (Fig. 3.3A)
and sites (Fig. 3.3B), with however mangrove relatively well separated from reefs (Fig. 3.3A
and 3.3B).
59
For both multi-species and L. fulviflamma analyses, mangrove was principally characterized
by Mn, Mg and Ba whereas fringing, intermediate and inner barrier reefs constituted three
close groups predominantly characterized by Sr (Fig. 3.2C and 3.3C). Importantly, however,
the differences between mangroves and reefs were clearer when using mono-specific data
(Fig. 3.3B) than multi-specific data (Fig. 3.2B).
Table 3.3: Multivariate results of PERMANOVA examining spatial variation in otolith
elemental composition within site and habitat for multi-specific and Lutjanus fulviflamma
(*** p < 0.001; ** p < 0.01; * p < 0.05).
A
Species
Df
MS
F
p (>F)
Multi-specific
Site
Habitat
Site x habitat
10
3
16
0.003065
0.075786
0.003374
1.994
49.295
2.194
0.006**
0.001***
0.001***
L. fulviflamma
Site
Habitat
Site x habitat
10
3
14
0.014207
0.0049035
0.0008362
3.2702
11.2864
1.9247
0.001***
0.001***
0.013*
B
C
Figure 3.2: Plots of principal component analysis (PCA) of elemental compositions of multispecific otoliths for habitats (A) and stations (B) and plot of contributions of elements in
habitat and stations elemental compositions (C) (M: mangrove; FR: fringing reef; IR:
intermediate reef; BR: inner barrier reef).
60
A
B
C
Figure 3.3: Plots of principal component analysis (PCA) of elemental compositions of L.
fulviflamma otoliths for habitats (A) and stations (B) and plot of contributions of elements in
intermediate reef; BR: inner barrier reef).
Results of PERMANOVAs and PCAs obtained with the remaining seven species were mixed
and are detailed in appendix (see Appendix 1 and 2). Differences between sites and habitats
were generally less significant or even non significant when considering species collected
only in reefs habitat (D. aruanus, C. lunulatus, C. striatus and G. aureolineatus) except for S.
bilineata. Concerning species collected only in mangroves (G. oyena and S. lineatus),
differences between stations were highly significant.
3.2. Accuracy of classifiers at different spatial scales
At the local scale (stations of the west coast sites), correct classifications of mangroves,
fringing, intermediate and barrier reef habitats using multi-species otolith datasets showed
moderate levels ranging between 49 and 69% depending on sites (Tab. 3.4). At the species
level, rates were much higher. For L. fulviflamma, correct classification rates were high,
ranging between 73 and 78% for all sites except Ouano where rate was only 43%. For D.
aruanus correct classification rates were even higher, ranging between 74 and 85%, except at
Prony where it reached 67%. For C. lunulatus rate of correct classification was 80% at Ouano
and StVincent and 67% at Gatope and Prony. Percentages of correct classification were the
highest for C. striatus with an average of 82% and a range of 71 - 96% depending on sites.
There was no sufficient data to conduct these tests for the other species at the local scale.
61
Table 3.4: Summary of RF classification accuracies in percent of individuals correctly
classified to the location they were collected. The highest classification accuracies (≥ 80%) are represented in bold and italic (Gat: Gatope, Oua: Ouano, Pro: Prony, StV: StVincent; M:
mangrove, FR: fringing reef, IR: intermediate reef, BR: barrier reef).
Species
Sample size (n)
Multi-specific
L.fulviflamma
D.aruanus
C.striatus
C.lunulatus
S.lineatus
G.oyena
S.bilineata
G.aureolineatus
M
FR IR
BR
Local
Habitat
M/FR/IR/BR
Gat Oua Pro
260
37
/
/
/
41
33
/
/
175
38
39
28
12
/
/
9
/
527
86
90
72
46
/
/
30
29
63
74
74
79
67
/
/
/
/
186
35
39
33
12
/
/
/
/
49
43
74
83
80
/
/
/
/
57
78
67
96
67
/
/
/
/
StV
Regional
Habitat
M/FR/IR/BR
4
69
73
85
71
80
/
/
/
/
59
67
64
71
67
/
/
95
/
Stations
M FR IR
4
4
4
63
100
/
/
/
84
83
/
/
42
63
64
71
50
/
/
78
/
44
54
67
64
58
/
/
/
/
BR
4
Global
Habitat Stations
M/BR M BR
11
11 11
47
48
60
63
83
/
/
100
80
90
95
/
/
/
/
/
/
/
At the regional scale (habitats of west coast sites), RF classification of the four habitats
resulted in low percentages of correct classification for the multi-specific analysis (59%),
ranging from 42% to 63% depending on the habitat (Tab. 3.4). For the four species tested,
mono-specific analyses showed higher percentages: L. fulviflamma (67% of individuals
correctly classified; range: 40% - 85% depending on the habitat), D. aruanus (64%; range
46% - 85%), C. striatus (71%; range 39% - 85%) and C. lunulatus (67%; range 25% - 91%)
(Tab. 3.4). Classification of sites within a specific habitat at regional scale showed similar
results. Mangrove stations showed a moderate correct classification rate for multi-specific
signatures (63%) but a very high rate at the species level, ranging from 83% to 100% for three
species (G. oyena, L. fulviflamma and S. lineatus). Individuals from fringing reef stations
were correctly classified in 42% of cases using multi-specific signatures, and 50% - 78 % of
cases with mono-specific signatures. Intermediate reef stations showed classification rates of
44% (multi-specific) and 54% - 67% (mono-specific). Correct classification of individuals
from barrier reefs was 47% (multi-specific) and ranged from 47% to 100 % depending on the
species (Tab. 3.4).
At the global scale (11 stations and 2 habitats around New-Caledonia), RF classifications
accuracy of mangroves versus barrier reef was extremely high using a multi-specific signature
(90% correct classification), and ranged from 81% in mangroves to 96% in reefs. At this
scale, data was sufficient to test mono-specific signature only for L fulviflamma. For this
62
44
65
/
/
/
63
67
/
/
29
29
41
47
41
/
/
37
24
species, 95% of individuals were correctly classified in the two habitats, with 86% correct
classification in mangroves and 99% in barrier reefs. Within the mangrove habitat, fish were
classified in the correct site in 44% of cases using the multi-specific signatures but in 63% 67% of cases using mono-specific datasets (G. oyena, L. fulviflamma and S. lineatus). Within
barrier reefs, correct classification of fish in their site of collection was poor for both multispecific (29%) and mono-specific signatures (24% - 47%) (Tab. 3.4).
3.3. Characterization of chemical signatures
Among the 62 classifiers built with the RF method, 14 were above or equal to 80% of correct
classifications (Appendix 3). B, Mn and Ba were the most recurrent element appearing in the
best classifiers. Among these and at the global scale, multi-specific otoliths collected in
mangrove and barrier reef of all stations were highly correctly classified (90%) with seven
elements constituting the best combination: Ba, Mg, Mn, Pb, Rb, Sn and U (Appendix 3).
Using only these elements, one-way multivariate PERMANOVA indicated a significant
difference between the two habitats illustrated in the PCA (Tab. 3.5 and Fig. 3.4). At the
regional scale, characterization of mangrove stations of the west coast sites with S. lineatus
otoliths is a good example of high correct classification (84%). For this signature, the best
combination of element was constituted of 2 elements: B and Mn (Appendix 3). Using these 2
elements, one-way multivariate PERMANOVA indicated a significant difference between
sites (Tab. 3.5) and PCA showed a clear discrimination between the four mangroves of the
west coast sites (Fig. 3.4). At the local scale, C. lunulatus otoliths collected in Ouano were
correctly classified in 80% of cases using five elements (Ba, Mg, Mn, Rb and Zn) (Appendix
3). One-way multivariate PERMANOVA of these 5 elements indicated a significant
difference between the three reefs illustrated in the PCA (Tab. 3.5 and Fig. 3.4).
elemental composition using elements from best combination only (Appendix 3) for multispecific at global scale, S. lineatus at regional scale and C. lunulatus at the local scale of
Ouano.
Species
Multi-specific
S. lineatus
C. lunulatus
Scale
Global
(11 sites)
Regional
(4 west coast sites)
Local
(Ouano)
Factor
Df
MS
F
p (>F)
Habitat
1
0.096
97.02
0.001***
Site
3
0.0006
20.261
0.001***
Habitat
2
0.0002
8.0877
0.004**
63
B
A
B
A
B
C. lunulatus – Local scale (Ouano)
S. lineatus – Regional scale (west coast sites)
M ulti-specific – Global scale (11 sites)
A
Figure 3.4: Plots of principal component analysis (PCA) of elemental compositions of
otoliths using elements from best combination only (Tab. 3.5) for habitats or stations (A) and
plot of contributions of elements in elemental compositions (B) (M: mangrove; FR: fringing
reef; IR: intermediate reef; BR: inner barrier reef; gat: Gatope, oua: Ouano; pro: Prony; stv: St
Vincent).
64
3.4. Multi-specific as a proxy of mono-specific
When multi-specific classifiers were used to predict habitat of single species, there was a
significant loss in correct classification compared to mono specific classifiers (Fig. 3.5,
Appendix 4). For instance, the rate of correct classification was 100% for L. fulviflamma
individuals collected in mangroves of the west coast sites using mono-specific signature, and
only 20% using global multi-specific signature. The only exception was observed at the
global scale between mangrove and barrier reef with a high correct classification using monospecific classifiers (95%) as well as using global multi-specific (90%) (Appendix 4). No
correlation was observed between the rate of correct classification with mono-specific and
multi-specific classifiers (R = 0.09, p > 0.5). This result remained when species were grouped
by diet (R = -0.39, p > 0.1) or family (R = -0.84, p = 0.01).
Figure 3.5: Plot representing correlations between percent of correct classifications obtained
with mono-specific classifiers and percent of correct RF classifications obtained with
corresponding global (blue circles), functional (red triangles) and taxonomic (green cross)
multi-specific classifiers.
65
4. Discussion
4.1. Characterization and spatial scales of chemical signatures in otoliths
In New Caledonian seascape, percentages of correct classification were higher at the habitat
level than at the station one. In fact, success in chemical distinction dropped down with the
number of stations which lead to an increase of the variability on the one hand and a decrease
in the number of replicates (individuals analyzed) on the other hand.
At the local scale, when habitats of west coast sites were taken individually, levels of
discrimination between habitats varied according to the species but were predominantly high
(>70% in most cases). Likewise, at the regional scale, when data of the four west coast sites
were pooled, elemental fingerprints of the four species tested showed a slightly reduced but
still high global level (>64%) of discrimination among habitats. Finally, at the global scale, L.
fulviflamma element fingerprints were clearly distinct between mangrove and barrier reef
(95% correct classification). At the regional and global scales, discrimination was lower at the
station level than at the habitat level. However, correct classifications of stations belonging to
the same habitat were quite high for mangrove species at the global level (>63%) and for both
reef and mangrove species at the regional level (>63% in half of the cases).
Contrasted distant habitats were discriminated with a high success because of the different
chemical influences from the coast to the barrier. A great discrimination was observed for
mangroves stations because otoliths from mangroves presented a higher chemical
heterogeneity than the reefs, mainly due to high concentrations of Mn. Mangroves receive
higher and different terrestrial inputs such as urban or freshwater inputs from numerous
estuaries along the New Caledonian coast (related to the characteristics of the catchments).
Fringing, intermediate and barrier reefs were chemically characterized by the presence of Sr,
but they overlapped and were either moderately spatially discriminated or misclassified.
These results could be explained by the spatial distribution of reefs habitats. Depending on the
site, fringing reefs are spatially closer to mangrove being under a coastal influence and
showing moderate heterogeneity of environmental chemistry. Terrestrial influences leaded to
distinguish fringing reefs from intermediate and barrier reefs, whereas intermediate reefs
close to barrier reefs were misclassified as barrier reefs. In others sites, fringing reefs were not
distinct from intermediate and barrier reefs because distance to the coast was higher and they
66
were under an oceanic influence through channels and passage in the barrier reef, leading to a
more homogeneous chemical composition of the otoliths.
Likely, integration of the spatial distribution of habitats at local scale may improve the power
of otolith fingerprints of reef habitats. For example some fringing and intermediate reefs can
be pooled in an inshore reef category, and some intermediate and barrier reefs in an offshore
category.
4.2. Inter-specific variations and multi-specific signatures
The present study also focused on the inter-specific variations of elemental compositions, the
ability to use multi-specific classifiers as a proxy of mono-specific ones and to estimate if the
use of taxonomic or ecologically closed species could enhance the potential as a proxy of
mono-specific classifiers. The interest of such proxy is to compensate a lack of fish to build
mono-specific classifiers when the studied species are difficult to collect (conservations
issues, rarity, difficulty to sample, etc.). In the present study, the percent of correct
classification obtained with mono-specific classifiers varied between species. Because
elemental fingerprints varied between species, the use of global multi-specific signatures
amplified the variability of the chemical signal and resulted in a high rate of
misclassifications. This increasing chemical variation is reduced but remained important when
multi-specific signatures are constituted of taxonomic closed species or species sharing the
same diet. However, when chemical contrast among habitats is very high, as between
mangrove and barrier reef then multi-specific signatures may prove useful.
Otolith microchemistry is based on the general assumption that their elemental composition is
directly related to the physicochemical properties of the surrounding environment (Bath et al.
2000). Several exogenous factors are reported to affect element incorporation, such as water
elemental compositions, temperature and salinity (Sturrock et al. 2012). For a given location,
all species collected were not exposed to the same variations of exogenous factors because of
their different ecology (mobility, diet, etc.). The mobility of species should be taken into
account when studying spatial patterns of elemental fingerprints. For instance, at the site of
Ouano, the success of classification of individuals was better for the least mobile species (C.
striatus, C. lunulatus and D. aruanus) than with the mobile L. fulviflamma. For L.
67
fulviflamma, Ouano was the site where the discrimination between habitats was the lowest.
This may be explained by frequent movements between habitats which were revealed in an
acoustic study at the same location and for the same species (Chateau et al. 2012). Acoustic
telemetry revealed patterns of movements at a lower temporal scale than the study of
elemental fingerprints. These movements varied between fish and high level of movements
between reefs habitats were observed, even several forays in mangrove for adult fish.
Furthermore, Kaunda-Arara and Rose (2004b) showed in their study that L. fulviflamma could
move over 1.5 km, which in some sites such as Ouano is sufficient to move between several
habitats.
Unlike mobile species, species with high site fidelity and a small home-range may show lower
variations in chemical composition of otolith. These fish will not be confronted to a change in
chemical composition of ambient water during their movements (Chittaro & Hogan 2013). An
acoustic study in Kimbe Bay (New Britain) revealed that outside the spawning season, the
movements of C.striatus were limited, their feeding area being restricted to a mean maximum
diameter lower than 13 m (Claydon et al. 2012). Furthermore, Mellin (2007) showed that this
species showed no change in habitat use during their ontogeny in New Caledonia. D.aruanus
and C.lunulatus present an extremely limited home range (Sale 1971, Guillemot et al. 2011).
Nevertheless, even if a low mobility may enhance the success of discrimination, a sedentary
species will still face temporal variations of the chemical composition of the ambient
environment.
Besides exogenous factors, endogenous factors like diet, stress, growth rate or ontogenetic
stage are suspected to influence element incorporation in the otoliths (Marohn et al. 2009,
Sturrock et al. 2012). In the present study, the use of species with similar diet for multispecific classifiers was not significantly more efficient than using data of all species. This
result was unexpected because diet was suspected to be a major endogenous process involved
in inter-specific variation of elemental fingerprints. However, studies testing if the variations
in otolith chemistry could be influenced by diet showed contrasted results. A few studies
found that variations of some elements were related to the food items ingested (Kennedy et al.
2000, Buckel et al. 2004, Walther et al. 2010). Bucket et al. (2004) found a significant
difference in Ba and Sr ratios between juveniles Pomatomus saltatrix fed with invertebrate
and finfish prey. However, they did not found a difference in Mn, Mg and Ca ratios. Walther
68
(2010) revealed significant effects of food on elemental concentrations of Ba and Sr but could
not distinguish them from effects of growth rate and temperature. However, most studies
revealed no significant difference in elemental concentrations between contrasted diets
(Milton & Chenery 2001, Lin et al. 2007). For example, Marohn et al. (2009) found no
variation in the concentration of Mg, Mn, Sr and Ba for European eels under eight different
diets. In another study, Walther and Thorrold (2006) revealed that Sr and Ba concentrations in
otoliths of juvenile Fundulus heteroclitus clearly reflect ambient Sr and Ba concentrations of
the environment. In this study, the majority of multi-specific signatures were determined from
carnivorous species and used to predict carnivorous species. However, the carnivorous group
is an extended diet class composed of species feeding on diverse trophic levels and in
different proportions (different size of mobile invertebrates ingested alone or mixed with a
variable proportion of fish prey). The poor efficiency of carnivorous multi-specific signatures
may therefore also be explained by the broad carnivorous class used in this study.
Before their incorporation in the otoliths, trace elements from the surrounding water have to
pass by physiological barriers (blood plasma and endolymph). These processes are species
specific, poorly understood and may result in inconsistent patterns of element incorporation
among species (Campana 1999, Sturrock et al. 2012). In this study, prediction using multispecific signatures of species from same family were realized under the hypothesis that
influence of these processes could be reduced with taxonomically close species (Swearer et
al. 2003). This hypothesis was not supported by our results with higher levels of
misclassification when species were grouped by family compared with multi-specific
signatures using all species. This implies that a precise knowledge of species specific
physiological processes is required to fully understand the inter-specific differences of
element incorporation into otoliths prior to using multi-specific signatures (Hamer & Jenkins
2007).
5. Conclusion
In New Caledonia, spatial scales of chemical discrimination with otolith fingerprints could be
revealed between contrasted habitats (mangroves and reefs) at the scale of the island (global),
69
the west coast sites (regional) and within sites (local). Discrimination at the station level was
lower and mainly related to the level of chemical heterogeneity within the habitats.
The results of this study showed that using multi-specific signatures as a proxy of monospecific signature is generally not possible, even when grouping species by family or by diet.
However, when studying highly contrasted habitats such as mangrove and barrier reefs, then
multi-specific signatures can prove very useful. Prior to use multi-specific signatures as a
proxy, future studies should quantify the effects of exogenous factors such as the mobility of
species, and endogenous factors such as the physiological processes of elemental uptake, diet
influence, growth rate and ontogeny.
Funding: This study was supported by the ZONECO program, the Institute of Research for
the Development, the University of New Caledonia and the South Province of New
Caledonia.
Acknowledgments: We thank the staff of the Institute of Research for the Development, the
University of New Caledonia, the Research Institute of Environment and Livelihoods of
Charles Darwin University and the Pole of Ocean Spectrometry of the European University
Institute of the Sea. In particular, we express our gratitude to Gerard Mou-Tham, Joseph Baly,
and Miguel Clarque for their invaluable field and laboratory assistance, Françoise Foti and
Claire Bassoulet for their precious help with the LA-ICP-MS analyses. David Mouillot and
Lény Mercier from the University of Montpellier 2 are warmly thanked for their help in
understanding the random forests statistical procedure.
70
Le chapitre 3 a permis d’évaluer la précision de la classification spatiale des sites et habitats obtenue à partir de la composition élémentaire des otolithes de plusieurs espèces de
poissons.
Le pouvoir discriminant de l’outil microchimique est important à l’échelle des habitats, particulièrement lorsqu’ils présentent des environnements contrastés comme par exemple les mangroves versus les récifs barrières. Les récifs frangeants et intermédiaires étant moins
contrastés ou sous des influences terrigènes variables en fonction des sites, il en résulte une
discrimination moins efficace. Néanmoins, il devrait être possible d’augmenter la précision des signatures en tenant compte de la distribution spatiale de ces deux habitats au sein de
chaque site. La microchimie hétérogène au sein d’un habitat spécifique telle que celle constatée au sein des mangroves augmente le pouvoir de classification à l’échelle spatiale plus fine des stations.
Ce chapitre a également permis de tester l’utilisation des signatures multi-spécifiques
comme substitut des signatures mono-spécifiques. Les résultats obtenus indiquent que dans la
grande majorité des cas le multi-spécifique ne peut se substituer au mono-spécifique et ce
même s’il est composé d’espèces taxonomiquement ou écologiquement proches de l’espèce constituant le mono-spécifique. Les différences interspécifiques d’incorporation des éléments au sein des otolithes sont vraisemblablement à l’origine de ce résultat. Ceci implique que les signatures microchimiques ne peuvent être développées qu’à partir de l’espèce étudiée. Des exceptions peuvent néanmoins se présenter, notamment lorsque les habitats à discriminer sont
très contrastés. En effet, les mangroves et récifs barrières sont parfaitement discriminés par
des signatures multi-spécifiques. Cela signifie qu’il est possible d’étudier la connectivité entre ces deux habitats même pour des espèces pour lesquelles un échantillonnage insuffisant n’a pas permis de construire des signatures microchimiques mono-spécifiques.
L’existence de fortes variations interspécifiques dans la chimie des otolithes implique que
chaque espèce intègre les signaux environnementaux de manière singulière. Afin de mieux
comprendre l’outil microchimie des otolithes, il apparaît donc nécessaire d’évaluer les
interactions entre la microchimie des otolithes et de l’environnement. Ceci est l’objet du chapitre suivant.
71
72
Chapitre 4: Comparaisons entre les signatures multi-élémentaires
de l’environnement et des otolithes en Nouvelle-Calédonie
73
74
Chapitre 4 : Comparaison de l’environnement et des otolithes
Chapitre 4: Comparaisons entre les signatures multi-élémentaires
de l’environnement et des otolithes en Nouvelle-Calédonie
La microchimie des otolithes est basée sur le principe que la composition élémentaire des
otolithes reflète plus ou moins directement celle de l’environnement (Bath et al. 2000). Or,
depuis le début de l’utilisation de la microchimie des otolithes en tant qu’outil de mesure de la connectivité écologique, d’aide à l’identification des stocks et de reconstruction de l’histoire environnementale des poissons, il a été montré que différents facteurs auraient une influence
sur l’incorporation des éléments traces au sein des otolithes.
Ce chapitre s’insère dans l’axe méthodologique de la thèse et a pour objectif d’étudier la microchimie de l’environnement en Nouvelle-Calédonie et les corrélations avec la
microchimie des otolithes. Il s’agira notamment de :
1) Caractériser les variations spatiales de la microchimie de l’environnement à l’échelle des sites et habitats de la Nouvelle Calédonie.
2) Evaluer la précision des signatures microchimiques de l’environnement, notamment dans l’eau et les sédiments, et examiner les différences entre ces deux compartiments.
3) Etudier les corrélations entre la composition élémentaires des otolithes, de l’eau, des sédiments, la salinité et la température.
Ce travail sur les relations entre la microchimie des otolithes et de l’environnement est utile pour mieux comprendre les différences et les points communs entre les signatures
observées dans les otolithes de différentes espèces collectées dans les mêmes environnements.
75
Manuscrit B: Comparison of otoliths and environmental multi-elemental
signatures in the coral reefs and mangroves of New Caledonia
C. Paillon 1, 2, L. Vigliola 1, D. Parry 3, C. Bassoulet 4, O. Bruguier 5, L. Wantiez 2, 6
1
2
3
Charles Darwin University (CDU), Chemistry & Microbiology Unit (ECMU), Darwin, Northern Territory,
Australia
4
Université de Bretagne Occidentale (UBO) - CNRS - Ifremer, Pôle de Spectrométrie Océan (UMS3113),
Plouzané, France
5
Université Montpellier 2 (UM2), Géosciences Montpellier (UMR5243), Montpellier, France
6
Aquarium des Lagons, Noumea, New Caledonia
Abstract
Otolith chemistry is a method increasingly used to reconstruct environmental histories and to
track lifetime movements of fish. However, it requires knowledge of relationships between
the physicochemical properties of the environment and the elemental uptake into otoliths for
each species of interest. We examined environmental (seawater and sediment) and otolith
microchemistries of several species collected in mangroves and reefs around the main island
of New Caledonia. We attempted to characterize the spatial variation of environmental
microchemistry and to evaluate the level of spatial discrimination possible. We also tested the
relationships between the elemental compositions of the otoliths and the environment, and
with the variations of temperature and salinity. Our analyses indicated that sediment could not
be chemically characterized at any spatial level. In contrast, seawater microchemistry showed
76
a high rate of correct classification at the level of habitat but not at the station level, except
among mangrove stations because of their high chemical heterogeneity. Effects of
environmental chemistry, temperature and salinity on otolith chemistry were complex and
diverse depending on the species or the environmental factor considered. Ba and Sr showed
classical variations with some species confirming theirs potential as indicators of past
environments for these species, but this was consistent with all species. Individual influence
of exogenous factors on Mn and Cr uptake were not clearly identified. However, these
elements seem to be valuable indicators of coastal habitats used by fish. Our results indicated
that precautions should be taken before using otolith chemistry as proxies for environmental
conditions. Nevertheless, reconstruction of past habitats used by fish can still be realized.
Key words:
environmental microchemistry, otoliths microchemistry, LA-ICP-MS, trace
elements
77
1. Introduction
Otoliths are paired structures located in the inner ear of teleost fishes and are composed of
calcium carbonate crystals, primarily aragonite, deposited within a protein matrix. Otoliths
grow continuously throughout life with calcareous layers deposited daily on their external
surface. Through these deposits, trace elements from the surrounding environment are
incorporated into the otolith and are never reworked through time due to the metabolically
inert nature of otoliths. Thanks to these key properties, otoliths can archive the life-history
information of a fish with a great time resolution. And indeed, the elemental composition of
fish otolith has become a powerful tool to reconstruct life history movements of fish
(McCulloch et al. 2005, Milton et al. 2008, Mercier et al. 2012), to discriminate fish stocks
(Campana et al. 2000, Ferguson et al. 2011), to assess connectivity between populations
(Gillanders 2002, Cuveliers et al. 2010) and to evaluate the importance of nursery areas
(Gillanders & Kingsford 2000, Brown 2006b, Tournois et al. 2013).
However, the use of otolith chemistry as a toolbox for the study of fish ecology is based on
the principle that, to some extent, the chemical characteristics of the environment are
transferred to otoliths (Elsdon & Gillanders 2003a, Sturrock et al. 2012). Furthermore,
microchemistry of otolith not only can be used to study fish, but can also be used as an
environment recorder. Indeed, if the relationships between otolith elemental composition and
environmental conditions are known, then retrospective measures of environmental
characteristics (elemental composition of seawater and sediment, temperature, salinity) and
quality (events of metallic pollution) should be possible (Forrester & Swearer 2002, Elsdon &
Gillanders 2003b). This is particularly interesting as species with different ecology, growth
and longevity may provide complementary information on past environmental changes at
different scales notwithstanding the analysis of archeological otoliths that may provide
detailed information of the environment thousands of years ago (Carpenter et al. 2003,
Disspain et al. 2011, Mercier et al. 2012).
Thus, the relationships between elemental composition of otoliths and physicochemical
properties of ambient environment (seawater and sediment elemental composition,
temperature and salinity) crossed by fish not only needs to be elucidated to reconstruct past
changes in the environment but also to better understand the otolith microchemistry toolbox to
study fish ecology (Elsdon & Gillanders 2003a, Elsdon et al. 2008, Sturrock et al. 2012).
78
Multiple factors are known to influence the uptake of trace elements in otoliths. There are
both exogenous factors such as temperature (Elsdon & Gillanders 2002, Martin & Thorrold
2005), salinity (Elsdon & Gillanders 2002, Lin et al. 2007) and chemistry of surrounding
waters (Elsdon & Gillanders 2004, Martin et al. 2013), and endogenous factors like ontogeny
(De Pontual et al. 2003), somatic growth rate (Miller 2011) and diet (Buckel et al. 2004).
Furthermore, trace elements from ambient environment have to cross a series of biological
boundaries (gills, plasma and endolymph) before incorporation in otoliths. These poorly
known physiological pathways of incorporation may influence trace element concentrations in
otoliths. In fact, uptake of elements may be species-specific because species exposed to the
same environmental conditions may show inter-specific variations of otolith fingerprint
(Gillanders & Kingsford 2003, Swearer et al. 2003, Martin & Wuenschel 2006, Hamer &
Jenkins 2007). In addition, the relationship between otolith and water chemistry may be
influenced by the mobility of the fish between different locations at different times (Milton et
al. 2008). Variations in trace element concentrations between the environment and the otoliths
also depend on the elements considered (Bath et al. 2000, Milton & Chenery 2001, Elsdon &
Gillanders 2003b). Yet, studies assessing the influence of exogenous and endogenous factors
influencing elemental uptake generally focused on a limited number of trace elements like Sr,
Ba, Mg and Mn (Bath et al. 2000, Elsdon & Gillanders 2002, Martin & Thorrold 2005,
Martin & Wuenschel 2006). Consequently, no general rules can be applied (Elsdon &
Gillanders 2003b) and it is difficult to use otolith chemistry as an archive of the chemistry of
the ambient environment (Campana 1999).
The main objective of this study was to assess the relationships between the elemental
concentrations in otoliths and the elemental concentrations in the environment, the
temperature and the salinity for several coral reef and mangrove fish species of New
Caledonia and for multiple elements. First, we characterized the environmental and otolith
chemistries of habitats and sites. Second, we looked at correlations between salinity,
temperature and concentration of 12 trace elements in seawater, sediments and otoliths of 9
coral reef and mangrove fish species. Third, we compared the ability of environment and
otolith microchemistry to distinguish different habitats at different spatial scale. Then,
chemical fingerprints present in the environment were compared with those present in otoliths
in order to determine whether signals present in otoliths were a simple translation of signals
79
present in the environment or if major changes occurred due to exogenous and endogenous
factors.
2. Material and methods
New Caledonia is a French territory located in the South West Pacific, 1500 km east of
Australia (Fig. 4.1). The archipelago is composed of several small islands and a main island
with a surface area of 17 000 km2, 400km long and 60 km large at its maximum. Around the
main island, the 1744 km long barrier reef delimits the largest lagoon of the world (31 336
km2) (Andréfouët et al. 2009). The distance between the coast and the barrier reef varies from
1 to 65 km (Paris 1981) and some parts of the lagoon are registered on the Unesco heritage
list since 2008. New Caledonia tropical seascape is constituted of several habitats and
ecosystems such as mangroves, seagrasses, algal beds, bare soft bottoms and coral reefs
(fringing, intermediate and barrier).
Fish samplings were realized during three years (2009, 2010 and 2011) at eleven sites (Fig.
4.1 and Tab. 4.1). The first sampling took place during the austral summer 2009 in two sites
(S2 and S5) where mangrove and inner barrier reef were sampled. The second sampling has
been carried out during austral winter 2010 in eleven sites around the main island. Two
habitats (mangrove and inner barrier reef) were sampled in seven sites and four habitats
(mangrove, fringing, intermediate and inner barrier reefs) were sampled in four sites
designated as “west coast sites” (S2, S3, S5 and S11). The third sampling took place during
austral winter 2011 in the “west coast sites” within the same four habitats.
In reef habitats, large species were collected by spear fishing and small species were
anesthetized using clove oil and collected with handnets. In mangroves, gillnets were
deployed along mangrove maritime front at high tide and fish were collected at low tide.
Different mesh sizes were used in order to catch fish of different size and at different
ontogenetic stages. Immediately after collection, fishes were stocked in a freezer (at -20°C)
until dissection.
80
Figure 4.1: (A) Location of New Caledonia in the South West Pacific; (B) study sites: circles
and triangles indicate sites sampled with 2 and 4 habitats respectively; (C and D) details of
sites S4 and S2 (M: mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier
reef).
During the first and second samplings, a multi-specific collect has been carried out with
respectively 44 and 25 species sampled (Tab. 4.1). The third sampling focused on three target
species with contrasting diet and mobility level. Lutjanus fulviflamma (Lutjanidae) is a
carnivorous Indo-Pacific snapper living in coastal habitats such as mangroves as juvenile and
reefs as adults (Thollot 1992b, Carpenter & Niem 2001a). Ctenochaetus striatus
(Acanthuridae) is an herbivore Indo-Pacific surgeonfish with a limited mobility and a home
range restricted to a hundred square meters (Carpenter & Niem 2001b, Claydon et al. 2012).
Dascyllus aruanus (Pomacentridae) is a planktivore and sedentary Indo-Pacific damselfish
with a restricted home range to a few square meters (Sale 1971, Carpenter & Niem 2001a).
Three samples of seawater and two samples of sediment were collected in 2010 in each
station (Tab. 4.1). Samples were collected at low tide in mangrove and independently of the
81
tide in reefs. In each station, replicates of seawater and sediment samples were 10 meters
apart. Seawater was collected near the sea surface and 10 ml was filtered in a polypropylene
tube for chemical analyses using a PES membrane filter (Polyesthersulfone, porosity of 0.45
µm). To avoid development of aquatic organisms, seawater was immediately fixed with 0.2
ml of nitric acid (HNO3 (2%), Suprapur®) and stocked in the laboratory in the dark. Each
sediment replicate was constituted of 1 kg of surface sediment and immediately stocked in ice
and then in a freezer until lab processing. Fine-grained sediments (< 63 µm) were collected
under wet conditions using a 63 µm diameter sieve and ultrapure water. Mixtures of finegrained sediments and ultrapure water were placed in a sterilizer (100°C) to evaporate. Dry
fine-grained sediments were ground using an agate mortar and stocked in clean plastic bags,
in the dark and dry atmosphere. Temperature (temperature sensor, 0.1°C) and salinity
(automatic salinity refractometer, Atago®, 1‰) were measured three times in each station.
Table 4.1: Sample size and analyses realized (with M: mangrove; FR: fringing reef; IR:
intermediate reef; BR: inner barrier reef; B: Brest; D: Darwin; M: Montpellier).
Sampling
Year
n stations
n habitats
ICP-MS
n samples analyzed
Otolith
2009
2
2 (M, BR)
LA-ICP-MS
218 (B)
2010
7
4
2 (M, BR)
4 (M, FR, IR, BR)
LA-ICP-MS
122 (B) and 592 (D)
2011
4
4 (M, FR, IR, BR)
LA-ICP-MS
130(B) and 60 (D)
Seawater
2010
7
4
2 (M, BR)
4 (M, FR, IR, BR)
SB-ICP-MS
90 (M)
Sediment
2010
7
4
2 (M, BR)
4 (M, FR, IR, BR)
SB-ICP-MS
58 (M)
2.2. Otolith preparation and chemical analyses
All material used for otolith handling was previously decontaminated in 5% ultrapure nitric
acid bath (24h), rinsed three times with ultrapure water (18.2 MΩ), dried and stored in clean
plastic bags under a laminar flow hood (HEPA 100). At the lab, fish samples were unfrozen
and each individual was measured (Fork length, FL) to the nearest mm and weighted to the
nearest g. Paired sagittal otoliths were extracted using ceramic tweezers and ultrapure water.
82
The chemical signatures of sampling sites were characterized by analyzing the chemical
composition at the surface of otoliths. To remove organic waste of otoliths surface, otoliths
were cleaned following Warner et al (2005) method with a bath of 50/50 H2O2 (30%,
Suprapur®) and NaOH (0.1 mol.L-1, Suprapur®) solution during 1 hour, sonicated during the
last 5 min of the bath, rinsed 5 times with ultrapure water for 5 min, dried under a laminar
flow hood (HEPA 100) and stored in individual plastic vials until chemical analyses.
The elemental signatures in otoliths at species level were determined from the analysis of up
to five otoliths randomly selected in each sample for each species. There were enough
individuals in the samples to determine these mono-specific signatures for eight species:
Dascyllus
aruanus,
Ctenochaetus
striatus,
Chaetodon
lunulatus,
Gerres
oyena,
Gnathodenthex aureolineatus, Lutjanus fulviflamma, Scolopsis bilineata, Siganus lineatus.
However, samples from all 53 caught species were used to determine multi-specific signatures
by randomly selecting up to three otoliths in each sample and for each species.
Otoliths were chemically analyzed using Laser Ablation Inductively Coupled Plasma Mass
Spectrometry (LA-ICP-MS) at the Pôle de Spectrométrie Océan (Institut Universitaire
Européen de la Mer, Brest, France) using a Thermo Element 2 coupled to a laser 193 nm
CopexPro 102 Coherent, and at the Resources Institute of Environmental Livelihoods
(Charles Darwin University, Darwin, Australia) using a Agilent 7500ce coupled to a UP – 213
nm laser ablation system (Tab. 4.1). Both ICP-MS were operated at low resolution using
argon as the carrier gas. Both laser systems parameters were set on a 90 µm laser beam
diameter and a frequency of 5 Hz. Each analysis lasted 120 s including 30 s of background
and 90 s of ablation (laser activated). The following 20 isotopes were measured: 7Li,
11
B,
85
Rb, 88Sr, 95Mo, 111Cd, 117Sn, 138Ba, 208Pb, 232Th, 238U, 25Mg, 43Ca, 47Ti, 51V, 52Cr, 55Mn, 60Ni,
65
Cu and 66Zn.
One laser ablation on the otolith surface was done to reveal the latest elements incorporated
from the habitat where the fish was collected. To standardize the analyses, the ablation was
always done at the same location, at the tip of the post rostrum on the otolith posterior side.
Calcium was used as an internal standard to compensate possible variation due to difference
in quantity of material ablated. To correct instrument drift, an external standard (National
Institute of Standards and Technology, NIST, 612) was analyzed twice at the beginning and at
the end of each session and also every ten ablations.
83
Raw ICP-MS data (counts per second, cps) were cleaned of outliers defined as values greater
than three times the inter-quartile distance during both the blank and the ablation windows
(Tukey 1977, Longerich et al. 1996, Heinrich et al. 2003). Then, data were transformed in
elemental concentrations (parts per million, ppm) and limits of detection (LOD) were
calculated following Longerich et al. (1996). Elemental concentrations lower than the LOD
were set to zero. Elements selected for statistical analyses had to met the following two
criteria 1) elemental concentrations in otoliths had to be higher than the LOD 70% of the time
in at least one habitat or one site and 2) the coefficient of variation of elemental
concentrations in the NIST 612 had to be inferior than 10% (Chittaro et al. 2004, Chittaro et
al. 2006). The use of two different ICP-MS and the time intervals between two sessions
realized with the same ICP-MS might influence the absolute values of concentration. To
avoid the effects of possible variations, elemental concentrations were transformed in percent
of all measured elements.
2.3. Sediment and seawater preparation and chemical analyses
Prior to analyses, successive acid digestions were realized in Teflon® beakers to dissolve
sediment (Totland et al. 1992). For each sample, 100 mg of fine sediment were digested twice
with strong acid attacks (Suprapur® grade HNO3 at 69% and HF at 40%) for 12 hours. Then,
the sediment was digested twice with moderate acid attacks using only HNO3 (69%) for 12
hours. Sediment and acid mixture was evaporated on hotplate at 130°C after each digestion. A
take-back was realized and diluted with ultrapure water and a 1/10 ml aliquot was collected.
Both seawater and dissolved sediment were diluted in a final solution with a salinity of 1‰ in
a matrix of 2.5% of HNO3 and 1.0% of Indium for solution-based SB-ICP-MS. To achieve
this, 0.25 ml of seawater sample was diluted with 9.4 ml of ultrapure water, 0.25 ml of HNO 3
and 1 ml of Indium. The same dilution was realized with sediment solutions. Indium was used
as an internal standard to correct instrument drift.
All environmental samples were analyzed at the Laboratoire Géosciences (Université
Montpellier II) using a SB-ICP-MS Agilent 7700x. Elements measured in seawater and
sediments were the same than in otoliths; as for otoliths, absolute concentrations in ppm (parts
per million) were transformed in percent of all measured elements.
84
2.4. Statistical analyses
All statistical analyses were performed using R statistical software (R Development Core
Team 2011). To describe spatial variation in elemental composition of environmental
samples, two-way (site x habitat) PERMANOVA (Legendre & Anderson 1999, Anderson
2001) with 999 permutations and principal component analysis (PCA) were performed.
Multivariate correlations between temperature, salinity, the chemical composition of the
environment (seawater and sediment) and the otoliths were evaluated using Mantel correlation
tests. Then, Spearman rank correlation tests were used to evaluate the relationships for each
element.
Random Forests (RF) algorithms were performed to compare the spatial scales (site, habitat or
station) of chemical signatures obtained from otolith and environmental elemental
compositions. RF classification method allows freedom from normality and homoscedasticity
assumptions (Breiman 2001) and is one of the most powerful methods for classification of
habitats using otolith chemical signature (Mercier et al. 2011). With the RF classification
method, two-third of the data set is used to build a classification tree and the remaining onethird is classified along this tree. The RF algorithm builds 5000 trees to ensure that every
sample gets predicted several times which permits to estimate classification accuracy (percent
of correct classification). All possible element combinations were tested to find the
combination that showed the highest percentage of correct classification. When several best
combinations were found, the one with the smaller number of elements was selected.
RF was performed at different spatial scales and for several organizational levels. Three
spatial scales were considered; local, regional and global. The local scale focused on
discriminations between habitats (mangrove, fringing, intermediate and barrier reef) within
each site of the west coast. The regional scale focused on the west coast sites together and
discriminations were done between habitats (mangrove, fringing, intermediate and barrier
reef) and between stations of a specific habitat. The global scale focused on the 11 sites of the
main island and discriminations were done at the level of habitat (mangrove versus barrier
reef) and between stations of a specific habitat. Elemental signatures were explored at the
species level for eight species (mono-specific signatures), at the multi-species level from 53
species and at the environmental levels on seawater and on sediment.
85
3. Results
In otoliths, twelve elements met our selection criteria and were thus used in the analyses: B,
Ba, Cr, Mg, Mn, Pb, Rb, Sn, Sr, Th, U and Zn. In seawater and sediments analyses, all
elements were always above the LOD but in order to compare chemistry of otoliths and
environment, the same twelve elements were retained in the environmental data sets. Two
samples from Amos seawater samples (S7) were removed from the data-set for all analyses
because values of Zn were extreme, revealing a possible contamination of the sample.
3.1. Characterization of environmental chemistry
Temperature of seawater varied between 22 and 26.2°C and salinity varied between 20 and 38
on all stations sampled during the study (Tab. 4.2). Salinity was more variable in mangroves
than in reefs with a much greater range of values and CV (Tab. 4.2). This was less clear for
temperature with a CV only slightly higher in mangroves than in reefs. There was no
correlation between the two variables (R = -0.02, p = 0.928).
At the global scale, with temperatures data of mangroves and barrier reefs of the eleven sites,
two-way PERMANOVA showed a significant interacting factor site x habitat (p-value <
0.001). Temperatures were lower in mangroves than in barrier reefs within five sites and the
difference was not significant within the remaining sites. At the regional scale, between the
four types of habitat of the West coast sites, two-way PERMANOVA also showed a
significant interacting factors site x habitat (p-value < 0.001). However, there was no clear
pattern of temperature variation between the habitats and sites of the West coast. Concerning
salinity, the two way PERMANOVA performed at global and regional scale resulted in
significant interacting factors site x habitat (p-value < 0.001 for both). However, there was no
clear pattern among habitat and sites at both scales.
Table 4.2: Mean temperature (°C) and salinity (S) by habitat with SE: standard error, CV:
coefficient of variation, N: number of station, n: number of measures (M: mangrove, FR:
fringing reef, IR: intermediate reef, BR: barrier reef).
Temperature
Mean
SE
23.6
±1.0
23.7
±0.8
24.1
±0.6
25.3
±0.8
Range
22 - 25.3
22.5 - 24.7
23.3 - 25.6
23.8 - 26.2
N
8*
4
4
11
n
24*
12
12
33
Salinity
Mean
33.6
36.8
36.7
35.7
SE
CV
Range
N
n
±3.9 12 %
20 - 37 9*
27*
M
±0.6 2 %
36 - 38 4
12
FR
±0.6 2 %
36 - 38 4
12
IR
±1.0 3 %
34 - 37 9*
27*
BR
(*) Because of technical failures, temperature was not measured in the mangrove of three sites (S7, S9 and S10,
see fig. 4.1) and salinity was not measured in the mangrove and barrier reef of two sites (S9 and S10).
86
CV
4%
3%
3%
3%
Using seawater samples, two-way PERMANOVA showed non-significant interacting factor
site x habitat (p-value = 0.99). Sites showed no significant differences (p-value = 0.43)
whereas differences between habitats were significant (p-value < 0.01). Plots of PCA of
seawater elemental compositions helped refine the significant habitat effect with mangroves
clearly distinguished from the three reef types (Fig. 4.2A) and confirmed the lack of
differences between sites (Fig. 4.2 B). Mangrove was mostly characterized by Mn and Cr, and
to a lesser extent by Sn and Zn (Fig. 4.2 A and C). The three reefs were distinct from
mangrove through the presence of Rb. Fringing and intermediate reefs formed two close
groups and were mostly characterized by Mg. Barrier reefs were predominantly characterized
by the presence of Sr and U.
A
B
C
Figure 4.2: Plots of principal component analysis (PCA) of seawater elemental compositions
for habitats (A) and stations (B) and plot of trace element contributions in habitat and stations
elemental compositions (C) (M: mangrove; FR: fringing reef; IR: intermediate reef; BR: inner
barrier reef).
Using sediment samples, a two-way PERMANOVA showed no significant site x habitat
factor (p-value = 0.88). There were no significant differences between sites (p-value = 0.54)
and between habitats (p-value = 0.65). PCA confirmed that environmental chemistry of
habitats or sites could not be characterized using sediment samples (Fig. 4.3).
87
A
B
C
Figure 4.3: Plots of principal component analysis (PCA) of sediment elemental compositions
barrier reef).
3.2. Correlations between environment and otolith chemistry
Although seawater and sediment elemental compositions were not globally correlated with
temperature or salinity (non-significant Mantel tests in Tab. 4.3), several elements showed
significant relationships. In seawater, B and Pb were positively correlated and Cr, Mg, Rb and
Th were negatively correlated with temperature. Mg was positively correlated and B, Ba, Pb,
and U were negatively correlated with salinity. There were no significant correlations between
elements in sediment and temperature. B and Sr showed a significant positive relationship and
Mg a negative relationship between sediment and salinity. Although seawater elemental
composition was not globally correlated with sediment elemental composition (nonsignificant Mantel tests in Tab. 4.3), Sr was the only one element which showed a significant
positive correlation.
88
Table 4.3: Results of univariate (Spearman tests) and multivariate (Mantel tests) correlations
between temperature, salinity and chemistries of seawater and sediment (“***” p < 0.001;; “**” p < 0.01;; “*” p < 0.05;; “+” p < 0.1).
Water
Salinity
B
Ba
Cr
Mg
Mn
Pb
Rb
Sn
Sr
Th
U
Zn
All
Rs
-0.56
-0.56
-0.05
0.41
-0.23
-0.38
0.29
-0.32
0.02
-0.04
-0.54
0.07
0.10
p
**
**
NS
*
NS
*
NS
NS
NS
NS
**
NS
NS
Water
Temperature
Rs
0.35
0.20
-0.43
-0.38
-0.32
0.44
-0.40
0.22
0.28
-0.44
0.27
0.11
0.04
p
+
NS
*
+
NS
*
*
NS
NS
*
NS
NS
NS
Sediment
Salinity
Rs
0.32
-0.11
0.02
-0.35
-0.06
-0.15
-0.09
-0.05
0.34
-0.15
0.10
-0.09
-0.10
p
+
NS
NS
+
NS
NS
NS
NS
+
NS
NS
NS
NS
Sediment
Temperature
Rs
-0.06
-0.14
-0.14
-0.09
-0.15
-0.29
-0.18
-0.31
0.08
-0.24
-0.09
-0.23
-0.10
p
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
Water
Sediment
Rs
0.00
-0.07
0.25
-0.16
-0.07
-0.04
0.19
0.19
0.31
0.23
-0.20
0.06
-0.04
p
NS
NS
NS
NS
NS
NS
NS
NS
+
NS
NS
NS
NS
Concerning otolith chemistry, multivariate Mantel correlations were significant in several
cases but univariate Spearman correlations were not always consistent between species (Tab.
4.4, see Appendix 5 to 8). With temperature, Mantel correlations were significant at the multispecific and species levels (L. fulviflamma and G. aureolineatus). At multi-species and for
several species, a positive relationship for Pb, Sr, Rb and U and a negative relationship for Cr,
Mg, Mn and Zn were observed with temperature. With salinity, Mantel correlations were
significant with otolith chemical compositions of L. fulviflamma and C. striatus only. At the
element level, positive correlations were observed with Sr, Zn and U and negative correlations
were observed with B, Cr, Mg, Mn, Sn and Zn for multi-species and several species. Mantel
correlation tests between seawater and otolith chemistry were not significant except for G.
oyena. However, spearman rank tests revealed a positive correlation for Mn, Pb, Th and U
and a negative correlation for B, Sr and Th. Finally, significant Mantel correlation coefficients
were observed between sediment and otolith chemistry for multi-species, L. fulviflamma and
G. oyena. At the species levels, positive correlations were observed for Mn, Sn, Sr, Th and U
and negative correlations were observed for Ba, Pb and Rb.
89
between temperature, salinity, environmental chemistry and multi-specific otoliths (“***” p < 0.001;; “**” p < 0.01;; “*” p < 0.05;; “+” p < 0.1).
Otolith
Salinity
B
Ba
Cr
Mg
Mn
Pb
Rb
Sn
Sr
Th
U
Zn
All
Rs
-0.40
-0.31
-0.37
-0.38
-0.35
-0.27
0.10
-0.56
0.42
0.07
-0.24
-0.41
0.11
p
*
NS
*
*
+
NS
NS
**
*
NS
NS
*
NS
Otolith
Temperature
Rs
-0.20
0.18
-0.62
-0.49
-0.34
0.26
0.29
0.09
0.39
0.06
0.50
-0.46
0.16
p
NS
NS
***
**
+
NS
NS
NS
*
NS
**
*
+
Otolith
Seawater
Rs
0.10
0.23
0.27
-0.19
0.85
0.36
-0.43
0.11
-0.12
0.05
0.08
0.19
0.16
p
NS
NS
NS
NS
***
+
*
NS
NS
NS
NS
NS
NS
Otolith
Sediment
Rs
-0.25
-0.20
0.05
-0.16
-0.08
-0.07
-0.27
0.12
-0.10
0.09
-0.06
-0.01
0.46
p
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
*
3.3. Accuracy of environmental and otolith classifiers
At the local scale, the percentage of correct classification of 4 habitats within west coast sites
using seawater dataset ranged between 50 and 79 % (Tab. 4.5). Prony showed the lowest
percentage of correct classification (50%), and Ouano presented a moderate correct
classification (67%) and Gatope (79%) and St Vincent (71%) showed good percentages of
correct classifications. For the sediment, percentages of correct classification were low for St
Vincent (43%) and Prony (50%), moderate for Gatope (63%) and high for Ouano (88%). In
contrast, percentages of correct classification based on otolith data were high (>70% in most
cases) for all species at all sites, except for L. fulviflamma and all species combined at the site
of Ouano (Tab. 4.5).
At the regional scale, when pooling data of the 4 west coast sites within each habitat, seawater
dataset permitted a high rate of correct classification (74%) of the four habitats (mangrove,
fringing, intermediate and barrier reef) whereas classification accuracy based on sediment
dataset remained low (42%). Similar to seawater, the analysis of elemental fingerprints of the
four species tested showed a relatively high level (>64%) of discrimination among habitats at
this scale. However, percentage of correct classification was lower (59%) when otoliths from
many species were pooled in the RF analyses (Tab. 4.5). Within mangroves, RF analysis
90
showed an extremely high rate of correct classification (86%) of the four sites using seawater
dataset. On the opposite, sediment samples were highly misclassified with only 25% of
correct re-assignment. Within fringing reefs, both seawater and sediment showed a low rate of
correct classification of the four sites (33% and 50% respectively). Within intermediate reefs,
seawater presented a low rate of correct classification of the four stations (50%) unlike
sediment (75%). Within barrier reefs, the four stations were misclassified both with seawater
and sediment (37% and 57%). As for seawater, percentages of correct classification using
otolith data were high in mangroves (83-100% depending on species). For reefs, accuracy of
classification varied among species; it was high for S. bilineata (78%) and C. striatus (71%)
in fringing reefs, D. aruanus (67%) in intermediate reefs, and S. bilineata (100%), G.
aureolineatus (80%) and C. lunulatus (83%) in barrier reefs.
Table 4.5: Summary of RF classification accuracies in percent of environmental samples and
otoliths correctly classified to the location they were collected. The highest classification
accuracies (≥ 80%) are represented in bold and italic (Gat: Gatope, Oua: Ouano, Pro: Prony,
StV: StVincent; M: mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier
reef).
Sample
Sample size (n)
M FR IR BR
Seawater
57 12 12 41
Sediment
22 8
8
22
Multi-specific
260 175 186 527
L.fulviflamma
37 38 35 86
D.aruanus
/
39 39 90
C.striatus
/
28 33 72
C.lunulatus
/
12 12 46
S.lineatus
41 /
/
/
G.oyena
33 /
/
/
S.bilineata
/
9
/
30
G.aureolineatus /
/
/
29
Local
Gat
79
63
63
74
74
79
67
/
/
/
/
Oua
67
88
49
43
74
83
80
/
/
/
/
Pro
50
50
57
78
67
96
67
/
/
/
/
StV
71
43
69
73
85
71
80
/
/
/
/
Regional
M/FR/IR/BR
74
42
59
67
64
71
67
/
/
/
/
M
FR
86 33
25 50
63 42
100 63
/
64
/
71
/
50
84 /
83 /
/
78
/
/
IR
50
75
44
54
67
64
58
/
/
/
/
BR
35
57
47
48
60
63
83
/
/
100
80
Global
M/BR M
63
95
76
23
44
90
65
95
/
/
/
/
/
/
/
63
/
67
/
/
/
/
At the global scale (eleven sites around the main land of New Caledonia), accuracy of RF
classification of mangrove and barrier reef was extremely high with seawater (95%).
Sediment also presented a high rate of correct classifications (76%). Within mangroves, the
eleven stations were correctly classified with a moderate rate (63%) for seawater but not for
sediment (23%). Within barrier reefs, the 11 stations were highly misclassified with both
seawater and sediment (22% and 30% respectively). Similar to environmental datasets,
classification accuracy of mangroves and reefs were very high with otolith datasets, 90% with
91
BR
22
30
29
29
41
47
41
/
/
37
24
the multi-specific and 95% for the only species (L. fulviflamma) that was collected in both
habitats. Similar to seawater, otoliths showed moderate rate of correct classification of
mangrove stations (63-67% depending on species). As for environment, stations of barrier
reefs were highly misclassified using otolith data (24-47% depending on species).
3.4. Comparison of environmental and otolith signatures
Seawater and sediment signatures obtained with RF classification method were generally
constituted of fewer elements than otoliths signatures. Furthermore, elements constituting the
otoliths signatures for a given scale varied according to species (see Appendix 3 and 9).
Despite the fact that all elements constituting the otoliths signatures were not part of the
environmental signatures, there were several common elements between both signatures. For
instance, the S. bilineata otoliths signature obtained to classify the three reefs at the regional
scale was constituted of B, Mn, Pb and Th, which were also observed in the environmental
signatures (Tab. 4.6). Among the elements constituting the C. lunulatus otoliths signature
distinguishing the barrier reefs at the regional scale, Ba and Pb were also found in seawater
signature and Cr and Sr in sediment signature.
92
Table 4.6: Comparisons between elements constituting otoliths and environment best
combinations for the highest otoliths RF classification accuracies (correct classification >
80%). Elements contributing to both otolith and environmental signatures are in bold. (M:
mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier reef).
Specie
Scale
Habitat
Correct classification (%) Combination of elements
L. fulviflamma
Multi-specific
Seawater
Sediment
Global
Global
Global
Global
M/BR
M/BR
M/BR
M/BR
95
90
95
76
B Cr Mn Pb Rb Sn Th U
Ba Mg Mn Pb Rb Sn U
Mg Mn Zn
B Mg Sr
S. bilineata
Seawater
Sediment
Regional
Regional
Regional
FR/IR/BR
FR/IR/BR
FR/IR/BR
95
73
61
B Mn Pb Sr Th
B Cr Pb Th
B Mn U
L. fulviflamma
G. oyena
S. lineatus
Seawater
Sediment
Regional
Regional
Regional
Regional
Regional
M
M
M
M
M
100
83
84
86
25
Ba Cr Mg Pb Rb Sn Zn
B Sn Zn
B Mn
Cr Mg Th
Cr
C. lunulatus
G. aureolineatus
S. bilineata
Seawater
Sediment
Regional
Regional
Regional
Regional
Regional
BR
BR
BR
BR
BR
83
80
100
35
57
Ba Cr Mg Mn Pb Sr
Ba
Mg Mn
Ba Pb
B Cr Sr
C. lunulatus
C. striatus
Seawater
Sediment
Local Ouano
Local Ouano
Local Ouano
Local Ouano
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
80
83
67
83
Ba Mg Mn Rb Zn
B Rb Sn U
Mg Pb
C. striatus
Seawater
Sediment
Local Prony
Local Prony
Local Prony
FR/IR/BR
FR/IR/BR
FR/IR/BR
96
33
50
B Cr Pb Rb Sn Sr Th Zn
B Pb
U
D. aruanus
C. lunulatus
Seawater
Sediment
Local StVincent
Local StVincent
Local StVincent
Local StVincent
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
85
80
77
60
B Sn Sr Zn
Ba Mn Rb
Mn Pb
Pb
Cr
4. Discussion
Otolith chemistry is a powerful tool frequently used in ecology to determine the
environmental history of fish. The technique principally relies on the basic assumption that
elements from the ambient environment incorporate in the otolith microstructure so that
93
otolith chemistry reflects the physicochemical properties of the environment. However,
results of this study showed that otolith chemistry is more complex than a simple reflection of
the chemistry of the environment.
4.1. Characterization of environmental chemistry and spatial scales of discrimination
Within New Caledonian seascape, elemental compositions of environment and otoliths
showed similar trends of spatial structuration, nonetheless sediment generally exposed a less
powerful rate of correct classification than seawater. Discriminations were higher between
habitats when considering contrasted habitats. Mangroves and reefs were greatly
differentiated and this difference was consistent through all sites. Discriminations between
reefs habitats were lower because of their similar chemistries. Stations of the same habitat
were not clearly discriminated except for mangroves stations at the regional scale for otoliths,
seawater and sediment.
Mangrove was clearly distinguished from reefs at the local (within sites, 4 habitats), regional
(west coast sites, 4 habitats) and global scales (all sites, 2 habitats) with otolith and
environmental chemistries. These high discrimination levels could be explained by the
specific chemical characteristics of New Caledonian mangroves, particularly high
concentrations of Cr and Mn in seawater. Because of its natural configuration, the impacts of
intensive mining activities and urbanization contribute to enhance the high natural soil erosion
along the shoreline of New Caledonia. This leads to high terrigenous inputs with associated
metals (Ni, Cr, Co and Mn) from the catchment to the lagoon (Ambatsian et al. 1997).
Furthermore, numerous estuaries exist along the coast of New Caledonia, increasing the
terrestrial influence through freshwater inputs. Therefore, a gradient of Cr and Mn was
observed from the coast to the barrier reef, with higher concentrations in coastal habitats
under terrigenous influence and lower concentration in reefs habitats under an oceanic
influence.
The power of discrimination obtained from sediment chemistry was generally weaker than
seawater resulting in a less obvious spatial structuration. However, the principal component of
mangroves sediments were fine-grained sediments which have a high affinity with trace
metals whereas reef sediment showed low level of this sediment-bound metal (Marchand et
al. 2011). Indeed, mangroves could be chemically distinguished from reefs using chemistry of
94
sediment and constituted one of the highest level of discrimination obtained from sediment
elemental composition in the present study.
Fringing and intermediate reefs chemically overlapped and were either moderately spatially
discriminated or misclassified. Depending on their location along the west coast, these reefs
were either under a high terrestrial influence (high terrigenous inputs) or an oceanic one
(direct link to a passage in the barrier reef). When under terrestrial influence, fringing and
intermediate reefs were generally grouped together and could be distinguished from both
mangrove habitat and barrier reefs, which presented a clear oceanic signature. When under
oceanic influence, fringing and intermediate reefs were generally misclassified with barrier
reef.
Sr and U predominantly characterized seawater chemistry of barrier reefs. With otolith
chemistry, Sr was also the major element characterizing reef habitats. Barrier reefs are distant
from shore and under an oceanic influence with rapid renewal of waters (Migon et al. 2007).
They were clearly distinguished from other reefs habitats in two west coast sites but were
mixed with fringing and intermediate reefs in the two remaining sites. The distinction of
barrier reef with fringing and intermediate mainly depends on the influence exerting on these
last two (oceanic/terrestrial) and also depends on the distanceof barrier reef from the coast.
4.2. Correlations between environment and otoliths chemistry
Otoliths are not in direct contact with the surrounding water as elements have to pass three
main filters (brachial uptake, cellular transport and crystallization) before incorporating in the
otoliths. These filters may either concentrate or dilute elements, meaning that otoliths do not
directly reflects the elemental composition of surrounding waters (Campana & Thorrold
2001). Even, if otolith chemistry does not reflect the exact composition of surrounding waters,
it may show proportional values and act as a proxy of environmental chemistry. Otolith
chemistry would then allow reconstructing the past environment of fish, and trace metal
concentrations in the otoliths may be related to the environmental exposure history of fish to
contamination. Such a property would also allow using the otolith chemistry tool without the
necessity of establishing an otolith chemistry baseline, by sampling directly the environment
(Dorval et al. 2007). This would be a cost effective way to characterize otolith chemistry and
retrace lifetime movement. But if otolith chemistry does not act as a proxy of environmental
chemistry, then an otolith library must be established to adequately study life history of fish.
95
Furthermore, it would not be possible to use otoliths as potential indicators of past metallic
contamination.
Before using otoliths to reconstruct environmental histories or define migratory patterns, it is
therefore essential to validate the various factors influencing the uptake of trace elements.
First, the general assumption was that the deposit of trace element metals in otolith reflected
the proportions of dissolved elemental concentrations in the ambient waters (Bath et al. 2000).
Then, several studies tended to show that elemental composition of otoliths was not only a
function of water elemental composition, as salinity and temperature also could influence
element uptake (Bath et al. 2000, Reis-Santos et al. 2013) indicating that the elemental
composition of otolith was related to both physical and chemical properties of the surrounding
environment, commonly qualified as exogenous factors affecting elemental incorporation
(Walther et al. 2010).
In the present study, complex and diverse relationships between environment and otolith were
observed, depending on the species or the environmental factor considered.
Concentrations of Ba and Sr in otolith generally showed the same relationships with
environmental factors as indicated in previous studies. Sr and Ba are known to be
incorporated proportionally to their concentration in the water (reviewed in Elsdon et al.
(2008)). In fact, concentrations of Ba in otoliths (multi-specific and L. fulviflamma) showed a
positive relationship with Ba in water, as observed in several studies (Elsdon & Gillanders
2002, 2003b, Martin & Thorrold 2005, Martin & Wuenschel 2006, Walther & Thorrold
2006). However, in the present study, Sr concentrations in otoliths did not show a significant
relationship with Sr concentrations in seawater unlike indicated in many studies (Bath et al.
2000, Milton & Chenery 2001, Elsdon & Gillanders 2003b). This may be explained by the
small range of salinity experienced in the present study since no estuary was sampled.
Ba concentrations in otoliths showed a positive relationship with temperature for only one
species (D. aruanus) confirming the inconsistent results of previous studies indicating
positive (Elsdon & Gillanders 2002, 2004) or non significant (Bath et al. 2000, Martin &
Thorrold 2005, Martin & Wuenschel 2006) relationships. Temperature has generally a
positive influence on concentrations of Sr in otoliths (Bath et al. 2000, Elsdon & Gillanders
96
2002, 2004, Martin & Wuenschel 2006), which was also found in our study with the multispecific group. The effects of salinity on Sr and Ba in otolith are generally described as
opposite with negative relationship on Ba (Bath et al. 2000, Elsdon & Gillanders 2002,
2003b, Martin & Thorrold 2005, Martin & Wuenschel 2006, Walther & Thorrold 2006) and
positive relationship on Sr (Bath et al. 2000, Elsdon & Gillanders 2003b, Walther & Thorrold
2006, Miller 2011). In the present study, salinity showed a negative relationship with Ba in
seawater but not in otolith as generally described. The positive relationship between salinity
and Sr in otolith was found in our study with the multi-specific group.
Mn and Cr presented a coastal to ocean gradient concentrations with high concentrations in
mangrove where temperature and salinity are slightly lower compared to reefs. The two
elements showed similar relationship between concentrations in otoliths and environmental
factors. A positive relationship between Mn in otoliths and in seawater was observed at the
multi-specific level and for L. fulviflamma and was also reported by Hamer et al. (2007) with
the otoliths of the snapper Pagrus auratus. However, non significant links were reported by
Elsdon and Gillanders (2003b) with otoliths of the Sparidae, Acanthopagrus butcheri. A
positive relationship was observed between Cr in L. fulviflamma otoliths and seawater. A
negative relationship between Mn in otoliths and temperature was observed with the multispecific group and the three species. This negative relationship was stronger for species
sampled in mangrove (multi-specific and L. fulviflamma). It was also found negative in the
literature (Miller 2009) or non significant (Elsdon & Gillanders 2002). A negative relationship
between Cr in otoliths and temperature was observed at the multi-specific level and for
L.fulviflamma, which is present in mangrove, and C. striatus, which is not present in
mangrove. Finally, a negative relationship with salinity was observed with Mn in multispecific otoliths only.
To our knowledge, no study focusing on factors affecting Cr incorporation in otolith was
published. The multiple relationships described in this study could be related to the higher
concentration of Mn and Cr in coastal habitats where temperature and salinity are lower.
Consequently, our results combined to a lack of literature on both elements did not allow us to
discriminate the effect of each factor on elemental incorporation and to identify which one
may drive the incorporation of Mn and Cr in otoliths.
97
In the present study, Mg in seawater showed a positive relationship with salinity whereas it
was negatively correlated with Mg in otoliths, the latter being also negatively correlated with
salinity and temperature. The negative relationship of Mg in otoliths with temperature was
observed at the multi-specific level and for L. fulviflamma and D. aruanus. The negative
relationship with salinity was observed only with the multi-specific group. Our results
contrasted with other studies focusing on exogenous factors driving Mg incorporation which
did not found significant correlation with salinity or temperature (Elsdon & Gillanders 2002,
Martin & Thorrold 2005, Dorval et al. 2007). Furthermore, Mg is not considered as a reliable
environmental indicator because it is suspected to be physiologically regulated (Martin &
Thorrold 2005, Hamer & Jenkins 2007, Miller 2011, Woodcock et al. 2012). Even if water
was proved to be the primary source of otolith Mg, concentrations in the otolith did not
change with its concentration in water (Dorval et al. 2007, Woodcock et al. 2012).
Zn is characteristic of anthropogenic inputs and can be a great indicator of past and actual
urban pollutions (Bruland et al. 2003). In the present study, a negative relationship between
Zn in otolith and temperature was observed at the multi-specific level and for D. aruanus. A
positive relationship between Zn in otolith and salinity was observed for D. aruanus. In
seawater, Zn follows nutrient-type distribution defined by relatively high concentrations in
coastal waters compared to oceanic waters masses (Bruland et al. 2003). In the present study,
Zn was not the major element distinguishing coastal habitat from offshore habitat but
participated to mangrove distinction even so. Therefore, Zn concentrations were higher in
mangrove habitat where temperature and salinity were lower compared to reef habitats. In
their study, Ranaldi and Gagnon (2008) did not rely Zn concentrations in water with
concentrations in otolith. In Thorrold et al. (1997) study, Zn concentration in otoliths was not
related to Zn concentration in ambient water. On the opposite, Arai et al. (2007) stated that Zn
appeared to be deposited in otoliths in relation to Zn concentration in the surrounding water.
Both studies retrospectively determined Zn concentrations throughout life of fishes, from the
core to the edge of otoliths. However, they did not possess retrospective data of temperature
and salinity of waters crossed by fish. Hence, these authors could not take into account the
possible influence of temperature and salinity on Zn uptake in otolith. Consequently, our
results combined to previous studies focusing on Zn uptake did not allow us to discriminate
individual effect of the three exogenous factors (water chemistry, temperature and salinity) on
Zn incorporation in the otolith.
98
Positives correlations between concentrations of Pb in otoliths of the multi-specific group and
of C. striatus were observed with temperature and with concentrations of Pb in seawater. The
last relationship was previously described under experimental conditions in two studies
(Geffen et al. 1998, Milton & Chenery 2001) where authors indicated that Pb concentrations
in otolith were related to water concentrations.
In the present study, positive correlations between concentrations of Rb in otolith for the
multi-specific group, L. fulviflamma and C. striatus were observed with temperature. On the
opposite, negative relationships between concentrations of Rb in otoliths of the multi-specific
group and of L. fulviflamma were observed with concentrations of Rb in seawater. A negative
relationship between Rb in seawater with temperature was observed. Rb presents a
conservative-type distribution in seawater, defined by concentrations that maintain a relatively
constant ratio to salinity (Bruland et al. 2003). To our knowledge, only one study focused on
variations of Rb concentration in otolith across an experimental salinity gradient (Hicks et al.
2010). The authors found a decrease of Rb concentrations in otoliths with salinity and no link
between Rb concentrations in otoliths and in water despite high Rb concentrations in
seawater. In the present study, we did not observe a relationship between salinity and
concentrations of Rb in otoliths. However, the range of salinity in the present study was much
smaller than in Hicks et al. (2010) study. Consequently, our results indicated complex
relationships of Rb in otolith and Rb in seawater, and with temperature. It is difficult to
identify which exogenous factors drive the uptake of Rb in otolith.
A negative relationship between concentrations of Sn in otoliths and temperature was
observed at the multi-specific level and L. fulviflamma and D. aruanus. The links between
concentration of Sn in otoliths and salinity were heterogeneous depending on the species
considered. Relationship with salinity was negative with Sn in L. fulviflamma otolith but was
positive with D. aruanus and C. striatus otoliths. Concentrations of B in otoliths showed a
few and mixed correlations with environmental settings. None of them were consistent
between species. Concentrations of Th and U in otoliths had no relationship with the
environmental factors. To our knowledge, no published study focused on the relation between
environmental settings and concentrations of Sn, B, Th and U in the otolith.
99
Relationships between elemental concentrations in sediment and elemental concentrations in
otoliths, temperature and salinity were weak. However, C. striatus showed recurrent
relationships between metal concentrations in otoliths and in sediment (positive with Mn, Mg
and Sr; negative with B). C. striatus is a micro-herbivore with a significant detritus
component in its diet. As it scrapes the sea bottom to collect microalgae the fish also ingests
surface layer with diatoms and detritus (Carassou et al. 2013). Relationships with sediment
chemistry may be related to its diet since it is known that food may have an effect on
elemental incorporation (Buckel et al. 2004).
Beyond exogenous factors, several studies highlighted the role of endogenous factors on
elemental uptake suspecting to be species-specific (Gillanders & Kingsford 2003, Swearer et
al. 2003, Hamer & Jenkins 2007). Diet (Buckel et al. 2004), somatic growth rate, genetic
(Barnes & Gillanders 2013), ontogenetic stage (Fowler et al. 1995) and even stress are
suspected to influence elemental incorporation in the otoliths (Marohn et al. 2009, Sturrock et
al. 2012). Endogenous factors are suspected to affect incorporation differently between
species and variations may exist within species, depending on the ontogenetic stage or the
somatic growth rate of individuals (Hamer & Jenkins 2007). In the present study, the effects
of endogenous factors could not be taken into account.
In New Caledonia, there is a set of elements that appeared to provide the best discriminatory
power for relating otolith metal concentration to the environmental history of the fish. Mn and
Cr representative of coastal area, concentrations in otoliths showed relationship with water
chemistry, temperature and salinity. We cannot identify separately which environmental
factors drive the uptake of metal in otoliths.
According to literature, Mg is subject to strong physiological regulation susceptible to be
specie specific (Martin & Thorrold 2005, Hamer & Jenkins 2007). The relationships of
exogenous factors with Mg in otoliths were inconsistent except with temperature, which may
be implicated in the physiological regulation. The relationships observed with concentrations
of Sn, B, Th and U in otoliths were too mixed, balancing between no consistency between
species (which could reflect an eventual specie-specific incorporation) or no relationship. For
these elements, there is a lack of bibliography to interpret the inconsistence as a speciespecific process of incorporation.
100
5. Conclusion
Otolith was considered by Bath (2000) as an excellent record of the physicochemical
properties of both present and past aquatic environment when focusing on Ba and Sr
variations. This hypothesis was reassessed by other studies with contrasting results and by
additional studies examining the influence of endogenous factors on elemental uptake (Buckel
et al. 2004, Barnes & Gillanders 2013). The mechanisms driving the relationship between
elemental concentration in otoliths and environment characteristics like water and sediment
elemental compositions, temperature and salinity, or with endogenous factors are complex
and not entirely understood (Martin & Wuenschel 2006). Consequently, precautions should be
taken when using otolith chemistry as proxies for environmental conditions (Walther et al.
2010).
The present study highlights the necessity to test several elements as environmental markers
and to not be limited on Ba and Sr, as elements like Mn and Cr are potentially indicative of
coastal waters and potentially reflect water metal concentrations. In order to understand the
driving mechanisms regulating elemental incorporation, there is a need to complete field
studies by experimental designs with fixed control conditions reflecting a natural range of
variations of environmental settings (temperature, salinity and water chemistry), along with a
control of endogenous factors with fixed diet, measured of somatic growth rate and
experimentation with individuals of the same species and ontogenetic stage (cohorts-specific).
The experimental and field study need to be done for several species with contrasting
ecological specificities (i.e. different diets) and taxonomically different (i.e. belonging to
different families).
Caledonia.
University of New Caledonia, the Geoscience Laboratory of Montpellier II University, the
101
Research Institute of Environment and Livelihoods of Charles Darwin University and the Pole
of Ocean Spectrometry of the European University Institute of the Sea. In particular, we
express our gratitude to Gerard Mou-Tham, Joseph Baly, and Miguel Clarque for their
invaluable field and laboratory assistance and to Chantal Douchet and Françoise Foti for
precious assitance with the ICP analyses.
102
Ce chapitre a permis de comparer le pouvoir discriminant des signatures microchimiques
présentes dans les otolithes de poisson et dans leur environnement.
La structuration spatiale de la microchimie de l’eau correspond à la structuration obtenue à partir des otolithes. Le pouvoir de discrimination spatiale est fort à l’échelle des habitats à partir de la microchimie de l’eau, particulièrement entre la mangrove et le récif barrière. A l’échelle des stations, la discrimination est faible excepté pour les mangroves où la
microchimie est plus hétérogène qu’au sein des récifs. Contrairement à l’eau, la discrimination spatiale à partir de la chimie des sédiments est généralement faible à l’échelle des sites et des habitats, excepté entre la mangrove et le récif barrière.
Les relations entre la microchimie des otolithes, celle de l’environnement et les paramètres physico-chimiques de la masse d’eau (température et salinité) sont complexes et ne sont pas toujours constantes entre les espèces. La chimie des otolithes n’est donc pas l’exact reflet de la chimie de l’eau car les otolithes et l’eau environnante ne sont pas en contact direct ;; des éléments franchissent plusieurs barrières physiologiques avant d’être incorporés à l’otolithe (Campana 1999). Toutefois, la forte structuration spatiale observée à
partir de la chimie de l’eau s’exprime de la même façon avec la chimie des otolithes. Ces résultats permettent l’utilisation logique de la microchimie des otolithes comme traceurs des environnements traversés par les poissons au cours de leur vie.
103
104
Chapitre 5 : Importance du rôle de nurserie des mangroves de
Nouvelle-Calédonie pour un poisson corallien, Lutjanus
fulviflamma
105
106
Chapitre 5 : Rôle de nurserie des mangroves
Chapitre 5 : Importance du rôle de nurserie des mangroves de
Nouvelle-Calédonie
pour
un
poisson
corallien,
Lutjanus
fulviflamma
La notion d’habitat juvénile ou « nourricerie » ou « nurserie » est essentielle à la
dynamique des populations marines (Beck et al. 2001, Adams et al. 2006, Dahlgren et al.
2006, Layman et al. 2006, Sheaves et al. 2006). L’efficacité d’un plan de conservation ne peut être optimal si les habitats juvéniles essentiels pour les espèces protégées ne sont pas
identifiés (Beck et al. 2001, Beck et al. 2003, Gillanders et al. 2003, Jones et al. 2010). Il est
donc crucial d’identifier les habitats juvéniles et d’estimer leur contribution au renouvellement des populations adultes afin de cerner les habitats à protéger en priorité
(Rooker et al. 2010).
La microchimie est un outil puissant couramment utilisé pour retracer l’utilisation des habitats au cours de la vie et donc pour l’identification des habitats juvéniles. Au cours de ce chapitre, le rôle de la mangrove comme habitat juvénile pour la dorade à tache noire Lutjanus
fulviflamma en Nouvelle Calédonie est évalué à travers :
1) La quantification de la proportion d’individus adultes ayant occupés une mangrove
comme habitat juvénile et l’estimation du temps passé au sein de cette mangrove.
2) L’évaluation de la distribution des mangroves calédoniennes en tant que déterminant
principal de la distribution de l’espèce dans l’archipel. Le travail méthodologique de caractérisation de la microchimie des otolithes réalisé dans
les chapitres précédents permet de révéler l’habitat juvénile de L. fulviflamma et de statuer sur
son caractère accessoire, essentiel ou obligatoire. Les données historiques sur l’abondance de l’espèce en Nouvelle-Calédonie combinées à la cartographie des mangroves permettent
ensuite d’évaluer l’influence de cet habitat juvénile sur la distribution des populations adultes.
107
Manuscrit C: Essentialness of mangroves as nursery habitat for a coral reef
fish in New Caledonia.
C. Paillon, 1, 2, L. Wantiez 2, 3, M. Kulbicki 4, M. Labonne 5, L. Vigliola 1
1
2
3
Aquarium des Lagons, Nouméa, New Caledonia
4
c/o Laboratoire Arago, Banyuls sur Mer, France
5
Institut de Recherche pour le Développement (IRD), LEMAR (UMR 6539), Institut Universitaire Européen de
la Mer, Plouzané, France
Abstract
For many fish species, mangroves of the Western Atlantic have long been recognized as
essential nurseries connected to coral reefs through ontogenetic migrations while there is still
a debate about the role of mangroves in the Indo-Pacific. The present study attempted to
estimate the importance of New Caledonian mangroves for juveniles of the blackspot snapper,
Lutjanus fulviflamma. Elemental composition of L. fulviflamma otoliths was used to
characterize the chemical signatures of mangrove and reef habitats of the main island of New
Caledonia, and to estimate the proportion of adults inhabiting mangroves as juvenile. Then,
the link between the distribution of L. fulviflamma and mangroves in New Caledonia was
tested using underwater visual census (UVC) and mangrove databases. Otolith elemental
108
compositions allowed a high percentage of discrimination between mangrove (83.8%;
characterized by Ba, Cr, Mn and Sn) and reef ecosystems (98.7%; characterized by Rb and
Sr). All L. fulviflamma adults analyzed presented a mangrove signature during their juvenile
stage with 75% inhabiting mangrove during the entire juvenile part of otolith analyzed.
Furthermore, the distributions of adult L. fulviflamma and mangroves were positively
correlated (r = 0.80 for occurrence, 0.91 for density and 0.86 for biomass). Importantly, L.
fulviflamma was absent from isolated islands of the New-caledonian archipelago where
mangroves were absent. These results emphasize the essentialness of mangrove in the life
cycle of L. fulviflamma in New Caledonia as an obligatory juvenile habitat for this species.
Key words: Lutjanus fulviflamma, otoliths, nursery, mangrove, coral reefs, UVC data
109
1. Introduction
Coral reefs and mangroves belong to the most diversified ecosystems of the world and have a
high economic value (Bellwood & Hughes 2001). They are also considered as deeply
impacted, with 30 % of coral reefs already severely impacted and a decrease of 50% of
mangrove original distribution (Hughes et al. 2003, Mumby et al. 2004, Burke et al. 2011).
Natural disturbances in conjunction with increasing human pressure lead to a dramatic loss of
biodiversity and habitats for these systems (Nystrom et al. 2000, Pandolfi et al. 2003,
Bellwood et al. 2004, Wantiez et al. 2006, Sandin et al. 2008, Mora & Sale 2011).
Coral reefs seascapes are fragmented ecosystems. These patchy structures imply a
fragmentation of reef organisms in local populations connected by movements of individuals
(Sale et al. 2010). Because most coral reefs species show a complex life cycle with a pelagic
larval phase and a benthic juvenile and adult phase, connectivity can be achieved by
movements of individuals at the larval, juvenile and/or adult stages. Typically, competent
larvae return to the reef environment and settle in shallow-water habitats like mangroves,
seagrass beds or reefs (Sale 1991). These shallow habitats may serve as nurseries where these
juveniles may benefit from several advantages such as food and shelter from predators
(Parrish 1989). As these fish grow, they move to coral reefs and recruit in the adult
populations (Mellin et al. 2007).
In order to protect biodiversity, Beck (2001) pointed up the necessity to include nursery
habitats in MPA networks. Identification of juvenile habitats for species with complex life
cycle is therefore critical for the management and conservation of species (Chittaro et al.
2009, Sale et al. 2010). In particular, the direct measure of connectivity between juvenile and
adult populations would allow managers to identify and protect the most productive nurseries
(Adams et al. 2006). This is important especially when these nurseries are vulnerable to
degradations and when connectivity between juvenile and adult habitats is threatened (Beck et
al. 2001, Brown 2006b), which is the case for the mangrove – coral reef system.
One type of movement connecting mangrove to coral reefs is ontogenetic migration. Many of
the coral reef species that use ontogenetic shift as life-history strategy use mangrove as
juvenile habitat (e.g. Serranidae, Lutjanidae, Lethrinidae, etc.) (Jones et al. 2010). Mangrove
110
are thought to be adequate juvenile habitat because they can provide shelter from predation
(complexity of roots, turbidity), supply food in abundance with an adapted size (zooplankton,
little crustacean and fishes), and therefore increase the overall rate of survival for juveniles
(Parrish 1989, Beck et al. 2001, Laegdsgaard & Johnson 2001, Jones et al. 2010). As a result,
there can be strong relationships between fish abundance and the presence of mangroves. For
example, numerous studies conducted in the Caribbean, showed greater biomasses of
Lutjanidae in reefs close to mangroves (Nagelkerken et al. 2000, Mumby et al. 2004,
Lugendo et al. 2007). For 17 species of Caribbean fish, Nagelkerken et al (2000) showed a
positive relationship between presence of mangrove and abundance on nearby reefs. For
regions lacking mangrove, 11 of the 17 species were not recorded or were seen in low
densities, highlighting the important role of mangroves in the life cycle of these species.
However, some species using mangrove during their juvenile life stage can also be observed
in high density in other type of juvenile habitats (seagrass beds, soft bottoms, fringing reefs)
(Mellin et al. 2007). Furthermore, Jones et al. (2010) showed that the absence of mangrove
may have no significant impact on adult fish population densities on nearby reefs. Thus,
mangroves may have a variable importance as nursery depending on the species and the
region. Some species may use mangrove as obligatory juvenile habitat if they cannot
complete their life cycle without mangroves. In this case, it is expected that these species are
absent when mangroves are absent. For other species, mangrove may simply be an essential
nursery habitat with a majority of juveniles using it. In that case, the species can be present
when mangroves are not but fish abundances should increase with mangrove extent. Finally,
mangrove may be an accessory nursery habitat with only a minority of juveniles found in
mangroves (Adams et al. 2006). In this case, no particular relationship should be expected
between mangrove extent and species abundance.
The connectivity between mangroves and reefs for ontogenetic shifters suggests that the
distribution of juvenile habitats may be a primary determinant of species distribution when the
juvenile habitat is an obligatory one. Because marine organisms disperse mostly by ocean
currents as larval propagules, it has often been assumed that larval transport was the primary
determinant of population connectivity and geographic range size. However, in a recent paper
Luiz et al. (2013) showed that adult traits were equal or better predictors of geographic range
size than larval traits, highlighting the importance of post-settlement processes likely to affect
111
the capacity of new colonizers to survive and establish reproductive populations. To some
extent, the study by Luiz et al. (2013) converges with the concept of “realized connectivity” (Hamilton et al. 2008), which takes into account the post-settlement fitness of individuals
with different larval traits. If larval and adult traits are important to the survival of individuals
after their movement to a new habitat, and ultimately to population connectivity and
geographic range size, then juvenile traits may also matters.
Testing whether the distribution of juvenile habitats is a primary determinant of species
distribution therefore requires determining if habitats such as mangroves are obligatory,
essential or accessories juvenile habitats. This implies back-calculating the proportion of
adults that effectively used a given habitat as nursery. Methods used to directly measure
connectivity between juvenile and adult populations require natural or artificial habitatspecific markers (Chittaro et al. 2004, Adams et al. 2006, Levin 2006). Among these
methods, microchemistry of otoliths is one of the most powerful tools to identify juvenile
habitats when adult and juvenile populations are spatially distinct (Gillanders & Kingsford
2000, Gillanders et al. 2003, Gillanders & Kingsford 2003, Gillanders 2005, Brown 2006b,
Jones et al. 2010). Otoliths are paired calcified structures located in the inner ear of fishes.
They show continuous growth allowing an incorporation of trace elements during the entire
lifetime of a fish. Due to their inorganic nature, trace elements are not resorbed after
incorporation. These key properties allow the use of otoliths as natural recorders of the
habitats inhabited by fish throughout their life (Campana 1999). In particular, it is possible to
analyze the chemical composition of the juvenile part of the otolith of an adult and
retrospectively determine the habitats used by this adult when it was a juvenile.
The purpose of this study is to 1) investigate whether, in New Caledonia, mangroves are
obligatory, essential or accessories juvenile habitats for Lutjanus fulviflamma (Forsskal, 1775)
(Lutjanidae) and 2) test whether the geographic distribution of mangroves in New Caledonia
could explain the adult distribution of this species. To achieve this, we used otolith
microchemistry to characterize the chemical signature of mangrove and reef habitats all
around the island and to estimate the proportion of adults which used mangrove as a juvenile
habitat. Then, we used an underwater visual census (UVC) database containing 2910
underwater transects to determine if the presence and/or the abundance of this species were
correlated with the presence and/or the abundance of mangroves.
112
2. Materials and methods
New Caledonia is an archipelago located in the South West Pacific, 1500 km east of
Australia, between latitudes 19°S and 23°S and longitudes 163°E and 168°E (Fig. 5.1). It is
composed of a main island (Grande Terre), three groups of smaller islands (Loyalties, Isle of
Pines and Belep) and some uninhabited islands and offshore reefs (Entrecasteaux,
Chesterfield, Astrolabe, Petri, Matthew, Hunter and Walpole). The main island holds the
largest lagoon of the world, with a linear barrier reef of 1744 km long and a surface area of 31
336 km2 (Andréfouët et al. 2009). Distance between the main island and the barrier reef varies
from 1 to 65 km (Paris 1981) and major parts of the lagoon are registered on the Unesco
heritage list.
sites in New Caledonia: circles indicate sites of otolith microchemistry study
Lutjanus fulviflamma (Forsskål, 1775) or blackspot snapper (Family Lutjanidae) is a small
Indo-Pacific snapper commonly targeted by subsistence fisheries. It is a carnivorous species
with a mean size of 25 cm, found in different habitats according to its ontogenetic stage.
Adults inhabit coral reefs between 3 to 35 m deep and juveniles are known to use shallowwaters habitats like mangrove (Loubens 1980, Thollot 1992a, Carpenter & Niem 2001a).
113
Fish were sampled for otolith analysis in different mangrove and reef habitats at 11 sites
around the main land of New Caledonia over three years (2009, 2010 and 2011) (Fig. 5.1).
Adults were collected in reef habitats by spear fishing and juveniles were collected in
mangroves using small mesh (7 mm) gillnets surrounding an isolated mangrove tree. Gillnets
were deployed at high tide and fish collected with handnets in a few cm of water and/or on the
ground at low tide. Immediately after collection, fishes were stored in a freezer (at -20°C)
until dissection. The first sampling took place during austral summer 2009 in the mangroves
and inner barrier reefs of two sites (Gatope (2) and St Vincent (5)). The second sampling has
been carried out during austral winter 2010 in the mangroves and inner barrier reefs of eleven
sites around the main island. During this sampling, fringing and intermediates reefs were also
sampled at four sites (Gatope (5), Ouano (3), Prony (11) and St Vincent (2)). Sampling in the
mangroves, fringing, intermediate and inner barrier reefs of these 4 sites was replicated during
austral winter 2011 (Tab. 5.1). Thus, our sampling included fish collected in different years,
seasons and habitats, which is important as chemical signatures stored in fish otoliths may
vary in space and time, and spatio-temporal variability needs therefore to be included in the
characterization for robust chemical signatures in otolith (Mercier et al. 2012).
Table 5.1: Sampling design in order to: A. characterize reefs and mangrove chemical
signatures by LA-ICPMS of the surface of otoliths and B. determine the juvenile habitat by
LA-ICPMS of transverse sections of otolith along transects from core to 1260 µm
A.
B.
Characterization of
habitat signatures
Determination of
juvenile habitats
Year
Site
Habitat
Laser analysis
2009
2
Mangrove and Reefs Ablation on surface
20
2010
11
114
2011
4
62
2010
4
Barrier Reef
20
Transects
n otoliths
2.2. Otolith preparation
All equipment used for otolith handling was previously decontaminated in 5% ultrapure nitric
acid bath during 24h, rinsed three times with ultrapure water (18.2 MΩ), dried under a
laminar flow hood (HEPA 100) and stored dry in plastic bag under the flow hood.
114
At the lab, fish samples were unfrozen and each individual was measured (Fork length, FL) to
the nearest mm and weighted to the nearest g. Paired sagittal otoliths were extracted using
ceramic tweezers, cleaned with ultrapure water (18.2 MΩ) and stored dry in plastic vials.
The chemical signatures of reefs and mangroves were characterized by analyzing the
chemical composition at the surface of otoliths of up to five fish randomly selected in each
sample (Tab. 5.1A). Otoliths of these fishes were deeply cleaned of organic material
following the Warner et al. (2005) method: otoliths were bathed in a 50/50 H2O2 (30%,
Suprapur®) and NaOH (Suprapur® 0.1 mol.L-1) solution during 1 hour, sonicated during the
last 5 min of the bath, rinsed 5 times with ultrapure water for 5 min, dried under a laminar
flow hood (HEPA 100) and stored dry in individual vials until microchemistry analyses.
The juvenile habitats actually used by adults were inferred from the analyses of chemical
composition along transverse otolith sections of 20 adult fishes collected on the barrier reefs
of Gatope, Ouano, Prony and St Vincent (5 fish per site, Tab. 1B). Transverse otolith sections
including otolith’s core were prepared by embedding right sagittal otoliths in resin (Araldite
2020®), sectioning transversely with a low speed saw (Buehler®), and polishing using
sandpaper of decreasing grain (from 800 to 1600 grains/cm2), lapping film (9 to 1 µm) and
diamond-tipped polishing powders (Struers®). Final sections were rinsed in ultrapure water,
dried under laminar flow hood (HEPA 100), mounted on a microscope slide (10 sections per
slide) with double-side tape and stored in plastic boxes until microchemistry analyses.
2.3. Otoliths LA-ICP-MS analyses
Otoliths were chemically analyzed by Laser Ablation Inductively Coupled Plasma Mass
Spectrometry (LA-ICP-MS) at the Pôle de Spectrométrie Océan (Institut Européen
Universitaire de la Mer, Brest, France) using a Thermo Element 2 coupled to a laser 193 nm
CopexPro 102 Coherent, and at the Research Institute of Environment and Livelihoods
(Charles Darwin University, Darwin, Australia) using an Agilent 7500ce coupled to a UP –
213 nm laser ablation system. Both ICP-MS were operated at low resolution using argon as
the carrier gas, and with the laser system parameters set on a 90 µm laser beam diameter and a
frequency of 5 Hz. Each analysis lasted 120 s including 30 s of background and 90 s of
115
ablation (laser activated). To discriminate habitats, the following isotopes were measured: 7Li,
11
B, 85Rb, 88Sr, 95Mo, 111Cd, 117Sn, 138Ba, 208Pb, 232Th, 238U, 25Mg, 43Ca, 47Ti, 51V, 52Cr, 55Mn,
60
Ni, 65Cu and 66Zn.
To access the latest microchemistry signature representative of the habitat where the fish was
caught, one laser ablation was done on the surface of the otolith of each subsampled fish (up
to 5 individuals per sample). To standardize the analyses, the ablation was always done at the
same spot at the tip of the post rostrum on the posterior side of otoliths.
To determine the juvenile habitat of adults, the juvenile part of otolith transverse sections
were analysed by LA-ICPMS for each of the 20 adult fish collected in the barrier reefs of
Gatope, Ouano, Prony and St Vincent. Because the otolith radius of the smallest juvenile
caught in our samples was 1260 µm, we considered this radius as the juvenile part of otolith
sections. For each adult, one LA-ICPMS transect was analysed. It consisted in ten successive
laser ablations (point-by-point mode) from the core to 1260 µm by increment of 140 µm
along the longest otolith growth axis.
To compensate possible variations due to differences in quantity of material ablated, calcium
was used as an internal standard. An external standard (National Institute of Standards and
Technology, NIST, 612) was analyzed twice at the beginning and at the end of each session (8
otoliths per session) and also after ten ablations for transect in order to correct instrument
drift.
Raw ICPMS data (counts per second, CPS) were first cleaned of outliers, defined as any value
greater than three times the inter-quartile distance, during both the blank and the ablation
windows (Tukey 1977). Cleaned CPS were then transformed in elemental concentrations
(parts per million, ppm) and limits of detection (LOD) calculated following Longerich et al
(1996). Only the elements that met the following two criteria were kept in statistical analyses:
1) elemental concentrations had to be greater than the LOD in 70% of the otoliths in at least
one habitat (e.g. mangrove) or one site (e.g. Ouano) and 2) the coefficient of variation of
concentrations in the NIST 612 had to be less than 10% (Chittaro et al. 2004, Chittaro et al.
2006). When the concentration of an element was under the LOD, it was set to zero. By
definition, this occurred in less than 30% of cases in at least one habitat or one site (otherwise
116
the element was removed from our analyses based on criterion 1). To reduce variation
originating from analyses produced by two different ICP-MS, elemental concentrations were
transformed in percent of all measured elements.
2.4. UVC surveys and mangrove area calculation
Fish diversity on the coral reefs of New Caledonia has been surveyed by IRD and UNC since
1986. Over the years, we built a database containing 2910 Distance-sampling Underwater
Visual Census (D-UVC). The method consists in recording species name, abundance, body
length and distance to transect line for each fish or group of fish observed along a 50 m
transect. Distance-sampling method is fully described in Labrosse et al. (2002).
Presence/absence of L. fulviflamma was directly extracted from our raw database and
abundance calculated after truncation of original D-UVC datasets at a distance of 5 m on each
side of the transect line (D’Agata et al. In press), effectively rendering D-UVC transects
equivalents to 500 m2 belt transects (50 m long x 5 m wide x 2 sides). Density of L.
fulviflamma was therefore calculated by dividing the number of fish counted within the
truncated D-UVC transects by 500 m2. Biomass was calculated from density and lengthweight coefficients (Kulbicki et al. 2005b). All transects were layed parallel to the reef
between 0 to 15 meters depth and encompassed the full range of habitats presents in New
Caledonia.
The presence, frequency of occurrence, average density and biomass of L. fulviflamma were
calculated for the different lagoons of New Caledonia (as defined by Pelletier (2012)) and the
remote reefs and islands of the archipelago for which we have data (Fig. 5.2). Surface of
mangroves standardized by coast length (to account for the difference in size of the different
lagoons) were calculated for the same lagoons, reefs and islands using the mangrove
cartography of Virly (2007) available at www.georep.nc.
117
Figure 5.2: Position of the Underwater Visual Census transects (UVC) in the archipelago of
New Caledonia. Polygons indicate the different lagoons, remote reefs and islands for which
we have data; numbers indicate the number of transects.
2.5. Data analyses
One-way PERMANOVA (with 999 permutations) based on Euclidian distance (Legendre &
Anderson 1999, Anderson 2001) was performed to test the effect of habitat (mangrove versus
reef) on the multi-elemental composition of otoliths. This test was followed by one-way
PERMANOVAs with 999 permutations in order to compare the chemical concentrations of
each element between mangroves and reefs. Non parametric PERMANOVAs were used since
the assumptions of normality and homogeneity of variance required by parametric
MANOVAs were not met.
Random forests (RF) algorithm (Breiman 2001) was performed on elemental concentrations
of otoliths surface to build an habitat classifier based on otolith microchemistry. This
classification method allows freedom from normality and homoscedasticity assumptions and
is the most appropriate for otolith microchemistry data (Mercier et al. 2011, Mercier et al.
2012). With the RF classification method, 75% of the dataset are used to build a classification
118
tree. Then, the remaining 25% are classified along this tree. The RF algorithm built 5000 trees
to ensure that every individual gets predicted several times, which permits to estimate
classification accuracy (percent of correct classification). Furthermore, we performed RF for
all possible element combinations in order to find the combination that showed the greatest
classification accuracy. The RF classifier based on this best combination of elements was then
used to predict the juvenile habitats corresponding to the chemical composition of the
different LA-ICPMS analyses performed along the otolith transverse sections of the 20 adults
collected on the barrier reefs of Gatope, Ouano, Prony and St Vincent. A PCA was also used
to further visualize the elemental composition of otoliths from fish collected in different
habitats. Elemental compositions were arcsin√x transformed in the PCA to normalize the
data.
The relationship between the presence of mangroves and L. fulviflamma in the different
lagoons, remote reefs and islands of the Archipelago of New Caledonia was tested by a Chisquare test. Spearman rank correlations were calculated to test the relationship between
mangrove and fish abundance (frequency of occurrence, density, and biomass).
3. Results
3.1. Building a RF habitat classifier from otolith microchemistry
Twelve elements met the criteria of selection and were retained for statistical analyses: B, Ba,
Cr, Mg, Mn, Sn, Sr, Rb, U and Zn. A PERMANOVA showed that the overall elemental
composition of otoliths was significantly different between mangrove and reef habitats (F
(1,194)
= 31.67; p < 0.01). One-way PERMANOVA further revealed significant differences
between habitats for six elements: Ba, Cr, Mn, Sn, Sr and Rb (Fig. 5.3). Elemental
concentrations of Ba, Cr, Mn and Sn were significantly in greater concentrations in otoliths of
individuals collected in mangrove than those collected in reefs. On the opposite, Rb and Sr
were in lower concentration in otoliths from mangrove than in reef otoliths.
119
F (1,194) = 11.09
F (1,194) =158.10
F (1,194) = 9.32
***
Cr (ppm)
Mn (ppm)
**
Ba (ppm)
**
Reef
Mangrove
Reef
F (1,194) = 8.74
Mangrove
Reef
F (1,194) = 13.65
F (1,194) = 31.97
***
***
Reef
Mangrove
Sr (ppm)
Sn (ppm)
Rb (ppm)
**
Mangrove
Reef
Mangrove
Reef
Mangrove
Figure 5.3: Otolith mean value of Ba:Ca, Cr:Ca, Mn:Ca, Rb:Ca, Sn:Ca and Sr:Ca ratios per
habitat, (R) reefs and (M) mangroves. Standard errors (± SE) are plotted. Results of one way
PERMANOVAs are indicated in graphs with (***) p<0.001 and (**) p<0.01.
The best Random Forest (RF) classification was obtained with a combination of only three
elements among the twelve tested: Mn, Sn and Rb. With these three elements, the RF was
able to correctly classify 95.9% of the fishes in the habitat where they were collected.
However, percentages of correct classification varied with the habitats. Fish from mangrove
were correctly classified in their sampling habitat in 83.8% and fish from reefs were correctly
classified in 98.7% (Tab. 5.2).
Table 5.2: RF classification matrix of habitats.
120
Habitats
Correct
classification (%)
n fish
classified in
mangrove
n fish
classified in
reef
n fish in
total
Mangrove
83.8
31
6
37
Reefs
98.7
2
157
159
Total
95.9
33
196
A PCA plot further showed the role of manganese (Mn), rubidium (Rb) and tin (Sn) in the
discrimination of mangrove from the reef habitats with Mn and Sn characteristic of
mangroves whereas Rb was characteristic of reefs (Fig. 5.3 and 5.4).
Figure 5.4: PCA biplot of individual fish collected in (M) mangroves and (R) reefs. Only
concentrations of Rb, Sn and Mn in otoliths were used in the PCA.
3.2. Back-calculating juvenile habitats of adults
Applying the RF best classifier to all the chemical analyses done from the core to the distance
of 1260 µm showed that all 20 adults L. fulviflamma collected on barrier reefs had a
mangrove signature on the two first laser ablations (0 and 140 µm) (Fig. 5.5). A mangrove
signature was revealed for 85 % of fishes (17 individuals) from 280 to the last analyses at
1260 µm.
121
100%
% of fish in a given habitat
90%
80%
70%
60%
50%
R
40%
M
30%
20%
10%
0%
0
140
280
420
560
700
840
980
1120 1260
Otolith radius (µm)
Figure 5.5: Prediction of habitats along the juvenile part of otolith transects of adults L.
fulvflamma collected in barrier reefs. Dark grey area represents fish in mangrove whereas
light grey area represents fish with reefs signatures (with M: mangrove and R: reef).
According to the classifier, only nly one fish used mangrove as juvenile habitat during a very
short period of time (Fig. 5.6). This fish was collected at the site of Prony.
Figure 5.6: prediction of juvenile habitat from an otolith transversal section of an adult from
Prony (site 11).
3.3. Relationship between mangrove and fish abundance
UVC surveys showed that the species was not recorded in places where mangroves are absent
(Chesterfield, Entrecasteaux, Astrolabe, Beautemps-Beaupré, Lifou) and was always recorded
in places where mangroves are present. Interestingly, some L. fulviflamma were recorded in
the atoll of Ouvéa where mangroves are present but not in the nearby atoll of Beautemps122
Beaupré where mangroves are absent. The strong relationship between the presence of
mangroves and L. fulviflamma on reefs was formalized by a highly significant chi-square test
(χ2=13, df=1, P<0.0003). Spearman rank correlations showed highly significant relationships
between standardized mangrove areas and the frequency of occurrence (R = 0.80, P<0.0009),
density (R=0.91, P<0.0001) and biomass (R=0.86, P<0.0001) of L. fulviflamma. Fish
abundance were 1) greater along the West coast of the mainland and in the NE lagoon where
mangroves are well developed, 2) lower in the North Lagoon where reefs are very far away
from the mainland, and the East, South East lagoons and Ouvéa island where mangrove are
smaller, and 3) absent in the reefs, atolls and islands disconnected from the mainland and
where mangroves are absent (Fig. 5.7).
Figure 5.7: Relationship between standardized mangrove area and density of L. fulviflamma
in the different lagoons, reefs and islands of the archipelago of New Caledonia (see map on
Fig. 2)
4. Discussion
The geographic distribution of species is a fundamental ecological parameter because it
strongly influences species susceptibility to extinction. As such, understanding the drivers of
123
species geographic distribution is crucial for the development of conservation strategic plans.
Up to now, it was believed that the main determinant of geographic range size of marine
organisms was larval dispersion and a few studies have indeed reported a positive relationship
between range size and planktonic larval duration (Mora et al. 2003, Lester & Ruttenberg
2005, Macpherson & Raventos 2006). However, many other studies found no relationship
(Thresher & Brothers 1989, Victor & Wellington 2000). Very recently, Luiz et al. (2013)
revised the larval dispersal paradigm and reported that adult traits were as important as larval
traits to explain the geographic range size of reef fish species. To explain their findings, they
proposed the following hypothesis: to found a new population, new colonizer not only must
arrive on a new habitat as settling larvae, but they must also survive and find a mate to
reproduce. And indeed, many adult traits such as large size or schooling behavior may
diminish predation risks and consequently increase the probability of new colonizer to survive
and establish new populations. If post-settlement survival is a crucial determinant of species
geographic range size, then we hypothesized that the spatial distribution of habitats, which are
obligatory for the survival of juveniles, should closely match the spatial distribution of
species. In order to test this hypothesis, we selected a reef fish species known to use
mangroves as nurseries, applied otolith microchemistry technique to determine whether
mangroves were obligatory, essential or accessory juvenile habitats, and explored the
relationship between species abundance and mangrove extent in the archipelago of New
Caledonia.
4.1. Characterization of juvenile habitat
The first objective of this study was therefore to identify the role of mangrove as juvenile
habitat for the blackspot snapper Lutjanus fulviflamma in New Caledonia using otolith
microchemistry. We found that the otolith chemical composition of Lutjanus fulviflamma
allowed discriminating almost perfectly mangrove from reef ecosystems. Six elements varied
significantly between the two ecosystems. Ba, Cr, Mn and Sn were in greater concentrations
in otoliths of fish from mangroves whereas Rb and Sr were in greater concentrations in
otoliths of fish from coral reefs.
New Caledonia shows a peculiar geology context and a significant past and present mining
activity. Among mining resources, past exploitations of Mn and Cr were found on the Main
Island. Mn was extracted mainly on the west coast and is still present in basaltic complex
124
grounds. Cr was intensely extracted on the south and the north-west coast (Paris 1981).
Moreover, Mn shows a scavenging profile type, which means higher concentrations in coastal
waters and a tendency to decrease with distance from coastline (Bruland et al. 2003, Sturrock
et al. 2012). Higher levels of Mn and Sn in mangrove could also originate from fertilizers
used for agriculture and could be higher near areas with freshwater inputs (Dove & Kingsford
1998, Chittaro et al. 2009). Furthermore, Sn was present in antifouling paints until late 2012
in New Caledonia, which could also have contaminated coastal waters near harbors. Finally,
differences in Ba concentrations between reefs and mangroves could be explained by Ba
concentrations inversely related to Sr concentrations in estuarine system (Sturrock et al. 2012)
where mangroves generally grow.
Our results show Sr in greater concentrations in otoliths from reefs and permit to discriminate
reefs habitats from mangrove. In others studies, patterns of Sr between habitats are variable.
In agreement with our study, Chittaro et al. (2004) reported Sr concentrations significantly
greater in otoliths from reefs than mangroves in Bahamas. Sr/Ca ratios have also been
positively correlated with salinity (Secor & Rooker 2000, De Pontual et al. 2003, Brown
2006a) and negatively correlated with temperature (Secor & Rooker 2000). However,
Yamashita (2000) found the opposite patterns: Sr/Ca ratios were negatively correlated with
salinity and positively correlated with temperature. Finally, Elsdon and Gillanders (2002)
found interactive effects of temperature and salinity on Sr/Ca ratios in otoliths with Sr/Ca
ratio negatively correlated with salinity, but that this effect was only present in medium
(20°C) and high (24°C) temperature treatments. Finally, Rb is a trace element with a
conservative profile, always in concentrations showing a constant ratio with salinity (Bruland
et al. 2003).
The power of otolith microchemistry to identify juvenile and adult habitats resides in distinct
chemical composition of otoliths collected from these different areas. In contrast to oceanic
waters, coastal waters often show chemical heterogeneity due to different inputs
(anthropogenic, fluvial, and upwelling) (Sturrock et al. 2012). Furthermore, coastal shallowwaters habitats are known to be important juvenile habitat (Parrish 1989, Beck et al. 2001). In
temperate environment, several studies reported specific habitat tags which permit to identify
juvenile habitat of fish (Forrester & Swearer 2002, Brown 2006a, Brown 2006b, Cuveliers et
al. 2010). On the opposite, water masses in coral reefs environment often showed limited
125
chemical contrasts. Generally due to a lack of sources of trace elements, studies failed to
detect strong differences among habitats at small scale (Chittaro et al. 2004, Chittaro et al.
2005, 2006). Nevertheless, some studies carried out at a small spatial scale in tropical
environment, showed strong chemical differences between habitats (Lo-Yat et al. 2005, Lara
et al. 2008, Mateo et al. 2010). In our study, lagoon habitats were discriminated from
mangrove habitats at relatively small spatial scale (2 to 20 km). This classification accuracy is
presumed to mainly derive from the environmental specificity of New Caledonia. In fact, the
Main Island gets several sources of traces elements from freshwaters and terrigenous inputs
linked to a singular geology context (Paris 1981). Thus, coastal shallow-waters habitats like
mangrove can show contrasted chemical composition with reefs habitats.
4.2. Essentialness of mangrove
The second objective of our study was to determine whether mangroves were obligatory,
essential or accessories juvenile habitats for L. fulviflamma in New Caledonia. To achieve this
goal, we looked for elemental signature of mangrove in the juvenile part of otoliths from
adults collected on barrier reefs. The results showed that all analyzed fishes had a mangrove
signature at the beginning of their life. However, the duration of juvenile life in mangroves
varied among individuals. Most fish inhabited mangrove during at least the entire juvenile
part of otolith (1260 µm), which corresponds to about one year of life. One individual,
however, used mangrove during a very short period of time, probably a few months during its
early juvenile phase. This pattern may be explained by the particular morphology of the site
where this fish was collected (Prony), an enclosed bay with fringing reefs adjacent to
mangrove areas. Distinction between elemental signatures of reefs and mangroves may be less
clear at this site.
Coastal shallow-waters habitats are recognized worldwide as important nursery areas for fish
(Parrish 1989, Beck et al. 2001, Adams et al. 2006). For a given species, the importance of
the nursery habitat may vary depending on the frequentation by juveniles. A habitat can be an
obligatory nursery if juvenile cannot survive without it. It can be an essential nursery if a
majority of juvenile uses this habitat and an accessory nursery when only a minority of
juveniles uses it (Adams et al. 2006). Many studies, mostly conducted in the Caribbean
region, established the nursery role of mangroves and their attractiveness for juveniles (Beck
et al. 2001, Laegdsgaard & Johnson 2001, Cocheret de la Morinière et al. 2003a, Mumby et
126
al. 2004, Dorenbosch et al. 2006, Lugendo et al. 2007, Nagelkerken et al. 2008, Jones et al.
2010). Microchemistry studies also showed the importance of mangroves as nurseries
(Chittaro et al. 2004, Chittaro et al. 2005, 2006, Mateo et al. 2010).
Our study took place in the South Pacific region where there is a lack of knowledge about
mangrove-reefs interactions in comparison with the Caribbean region (Adams et al. 2006). In
New Caledonia, previous studies showed that mangrove was used by blackspot snappers as
one juvenile habitat among other shallow-water habitats (Thollot 1992a, Carpenter & Niem
2001a, Igulu et al. 2011). Our study reveals the true importance of mangroves for L.
fulviflamma: all analyzed adults lived in mangroves as juveniles, which suggests that
mangrove is likely to be an obligatory juvenile habitat for this species in New Caledonia.
Although a sampling of 20 adults is probably not sufficient to affirm with certainty that
mangrove is indeed an obligatory habitat, the relationship that we found between the presence
of mangroves and the presence of the species strongly support our conclusion from otolith
microchemistry. New Caledonia is an archipelago with a large main island colonized by
mangroves and surrounded by a complex reef ecosystem, and several islands and reefs
offshore the mainland, some with mangroves and others without mangroves. L. fulviflamma
was present all around the mainland, including on reefs very far away from the land (the
northern and southern tips of the barrier reef), suggesting that individuals which completed
their juvenile life in the mangroves of the mainland were able to swim long distances within
an interconnected reef ecosystem and reach far away reefs. Furthermore, the species was
present in offshore islands where mangroves are present, suggesting that planktonic larvae
transported by oceanic currents can easily reach these islands and survive as juveniles in the
mangroves. However, the species was not present at d’Entrecasteaux atolls, Astrolabe reef,
Chesterfield banks, Beautemps-Beaupré Atoll and Lifou Island. One hypothesis to explain
this absence is a lack of connectivity at the larval stage. However, this hypothesis is very
improbable. For example, it would be very surprising that larvae can reach the atoll of Ouvéa
where the species is present and not the nearby atoll of Beautemps-Beaupré or Astrolabe reef
where the species is absent. Clearly, the most plausible explanation for the absence of the
species at these reefs and islands is the absence of mangrove, implying that indeed mangroves
are obligatory juvenile habitat for this species in New Caledonia. The strong relationship
between L. fulviflamma abundance, either in terms of frequency of occurrence, density or
127
biomass, and mangrove area further strengthens the importance of mangrove in the life cycle
of this species in New Caledonia.
Depending on the location, many species show considerable plasticity in habitat use as
juveniles (McMahon et al. 2012). In our study, juveniles of L. fulviflamma used mangroves as
an obligatory nursery area. In Zanzibar, Lugendo (2005) showed that juvenile of L.
fulviflamma were found in different studied embayment habitats (mangroves, sand/mud flats
and seagrass areas) and were therefore considered as a generalist species. In the Caribbean
region, Mateo (2010) showed that almost 100% of schoolmaster (Lutjanus apodus) collected
in mangrove and seagrass beds of St Croix and Porto Rico resided as juvenile in mangrove
habitat. Juvenile habitat could possibly be used at different levels in different regions
(Gillanders et al. 2003). This plasticity and variability in effectively used juvenile habitat
could be significant enough to limit the general applicability of our results (Adams et al.
2006). Nevertheless, our study demonstrates that the spatial distribution of L. fulviflamma in
New-Caledonia closely match the distribution of its juvenile habitat, which is in agreement
with the study by Luiz et al. (2013) which proposes that post-settlement processes are main
determinants of species geographic range. Our study encourages determining which habitats
are obligatory or essential for fish juveniles in the different regions of the world because these
habitats are probably crucial for the spatial distribution of species and would probably require
some protection, and/or restoration for species maintenance and conservation.
Caledonia.
University of New Caledonia, the Pole of Ocean Spectrometry of the European University
Institute of the Sea and the Research Institute of Environment and Livelihoods of Charles
Darwin University. In particular, we express our gratitude to Gerard Mou-Tham, Joseph Baly,
and Miguel Clarque for their invaluable field and laboratory assistance, Claire Bassoulet and
Françoise Foti for their precious help with the LA-ICP-MS analyses.
128
Le chapitre 5 a permis d’identifier le rôle de la mangrove comme habitat juvénile obligatoire pour Lutjanus fulviflamma en Nouvelle-Calédonie. La reconstruction par
microchimie des otolithes de l’histoire de vie environnementale d’adultes collectés sur le récif
barrière a démontré que l’ensemble des individus présentait une signature élémentaire caractéristique de la mangrove dans la partie juvénile de leur otolithe. De plus, la distribution
spatiale de l’espèce en Nouvelle-Calédonie s’est révélée être étroitement liée à la présence de mangrove.
Les mouvements ontogénétiques entre mangrove et récifs correspondent à un transfert
unidirectionnel d’individus des habitats juvéniles vers les habitats adultes (Berkstrom et al.
2012). Cependant, les modalités de ce transfert sont peu connues. Les L. fulviflamma juvéniles
quittent-ils les mangroves de manière définitive ou l’espèce y revient-elle à l’âge adulte ? La
migration de la mangrove vers le récif barrière est-elle directe ou s’effectue-t-elle via
l’occupation progressive d’habitats intermédiaires tels que les récifs frangeants et
intermédiaires ? Ces questions font l’objet du chapitre suivant.
129
130
Chapitre 6 : Connectivité et patrons de mouvement d’une espèce
de poisson corallien, Lutjanus fulviflamma
131
132
Chapitre 6 : Patrons de mouvement
Chapitre 6 : Connectivité et patrons de mouvement d’une espèce de poisson corallien, Lutjanus fulviflamma
Les habitats constitutifs des paysages récifaux-lagonaires forment une mosaïque de
patchs interconnectés par des processus écologiques et biophysiques mais menacés par une
pression humaine grandissante. La préservation de cette connectivité à travers tous les stades
de la vie d’un poisson constitue un objectif clé dans la conservation des écosystèmes et se
réalise notamment par la création de réseaux d’aires marines protégées (Almany et al. 2009,
Sale et al. 2010). La compréhension des patrons de mouvements entre les habitats constitutifs
des paysages récifaux-lagonaires est d’une importance primordiale pour le design de ces réseaux mais reste un phénomène mal caractérisé (Parrish 1989, Dorenbosch et al. 2005b,
Lugendo et al. 2005, Lugendo et al. 2006).
La microchimie des otolithes est actuellement la technique la plus performante pour
estimer la connectivité réelle entre les habitats à l’échelle de la vie entière d’un poisson (Campana 1999 , Gillanders 2005, Thorrold et al. 2007, Rooker et al. 2010). En effet, la
combinaison des études optiques et microchimiques des otolithes permet de déterminer
rétrospectivement l’histoire de vie environnementale des individus, de comprendre les patrons
de mouvements des poissons et d’étudier l’utilisation des habitats (Cowen et al. 2007, Elsdon
et al. 2008). La reconstruction de l’histoire de vie environnementale par microchimie des otolithes consiste à associer les changements de composition chimique des otolithes présents
le long de l’axe de croissance des otolithes avec les mouvements des poissons à travers
différents environnements (Campana 1999).
Dans ce chapitre, l’utilisation des habitats et les patrons de connectivité au cours de la vie
de L. fulviflamma sont étudiés via :
1) L’estimation du temps passé au sein des mangroves pour les juvéniles.
2) La caractérisation de la transition entre la mangrove et les récifs.
3) L’étude des patrons de mouvements au cours de la vie adulte.
133
Ce chapitre s’appuie sur le travail méthodologique de caractérisation de la microchimie
des otolithes réalisé en amont et qui permet d’associer les changements de composition
élémentaire avec les mouvements de ces poissons.
134
Manuscrit D: Migratory paths and connectivity for a coral reef fish, Lutjanus
fulviflamma.
C. Paillon, 1, 2, L. Vigliola 1, L. Wantiez 2, 3
1
2
3
Aquarium des Lagons, Nouméa, New Caledonia
Corresponding author: christelle.paillon@ird.fr
Abstract
The increasing pressure exerted on tropical coastal ecosystems such as mangroves and coral
reefs is leading to their degradation by threatening their resilience. A precise knowledge of
existing ecological connectivity within tropical seascapes can provide an important
contribution to enable the protection of these ecosystems. The present study used multielement otolith fingerprints of the Indo-Pacific blackspot snapper Lutjanus fulviflamma to
investigate the ecological connectivity between habitats of New Caledonia lagoon. We
identified lifetime patterns of movements between the different habitats, characterized habitat
use during juvenile and adult life stages, and described the transition between juvenile and
adult habitats. Fish were captured in mangrove and reefs within four sites of New Caledonia
Main Island. Using otolith elemental compositions and Random Forest classification method,
the chemical signatures of mangrove, inshore and offshore reef habitats were characterized
within each site with moderate to high accuracies (percentages of correct classification
ranging from 50% to 100%). Then, signatures were used to predict habitat use during the
lifetime of twenty L. fulviflamma adults. The prediction displayed a high level of global
135
accuracy (nearly half of total habitat predictions showed a probability ranging from 75% to
100%, the remaining half ranging from 50% to 75%). Reconstruction of environmental life
history revealed that all individuals occupied mangroves when juvenile. This mangrove stage
lasted at least a year, for a majority of the fish, to a maximum of 3 years. It also revealed three
different lifetime patterns of movements between habitats: one with a single adult reef habitat,
one with two adult reef habitats and one with multiple habitats use.
Key words:
connectivity, patterns of movements,
microchemistry, New Caledonia
136
Lutjanus
fulviflamma,
otolith
1. Introduction
Tropical seascapes are spatially heterogeneous appearing as mosaics of fragmented habitats
that include coral reefs, mangroves, seagrass beds, soft bottoms and several other habitats
such as flats or embayments. This patchy nature results in fragmented local populations of
marine organisms which are more-or-less connected (Almany et al. 2009). Connectivity
defined as the exchange of individuals among local populations (Cowen et al. 2006, Crooks &
Sanjayan 2006) can be measured in different ways (Calabrese & Fagan 2004). Structural
connectivity is a metric that includes only the physical properties of the tropical seascapes
while potential connectivity is a metric which embraces both physical properties and limited
knowledge on mobility of organism. The direct quantification of movement and dispersal of
organisms throughout the seascape refers to the metric of actual connectivity (Crooks &
Sanjayan 2006, Moilanen & Hanski 2006, Nagelkerken 2009). This metric is ecologically
highly relevant since it describes the real movement of organisms among local populations
but it is the most challenging to measure.
For coral reefs fish, larval dispersal has long been considered as the major if not unique event
of dispersion during life while juveniles and adults were considered as sedentary stages (Sale
1991). Yet, recent studies revealed that larval dispersal extent may be less than expected with
rates of self-recruitment reported as high as 60% (Almany et al. 2007). On the opposite,
dispersion by juvenile and adult movements may be more common than suggested (KaundaArara & Rose 2004a, Grüss et al. 2011). Exchanges of individuals among local populations or
connectivity at juvenile or adult stages have been classified in different types of movements
operating at different spatiotemporal scales (Berkstrom et al. 2012): tidal linked movements,
daily foraging and sheltering routine, spawning migrations and ontogenetic shifts. Through
these movements, numerous reef fish species utilize multiple habitats during their life.
Connectivity between habitats enhances the resilience of ecosystems constituting tropical
seascapes. The resilience of an ecosystem is defined as its capacity to absorb recurrent
disturbances, regenerate or adapt to change without permanently switching to an alternative
stable state (Mumby & Hastings 2008, Hughes et al. 2010). However, the continuous and
increasing human pressure exerting on coral reefs is responsible for degradation and loss of
habitats leading to a breakdown in connectivity within seascapes. This breakdown is
threatening resilience of coral reef ecosystems (Mumby & Hastings 2008). Likewise, not only
137
coral reefs are threatened in the tropical seascapes but mangroves are also currently facing a
high level of anthropogenic stress leading to a degradation and loss of these areas in the world
(Valiela et al. 2001, Polidoro et al. 2010).
The twin degradation of coral reefs and mangroves in most tropical seascapes may have
profound consequences for those numerous species that use mangrove as juvenile habitats and
coral reefs as adult habitats (Nagelkerken et al. 2002, Cocheret de la Morinière et al. 2003a,
Chittaro et al. 2004). In tropical seascapes, marine protected area (MPA) networks are
established to preserve and maintain connectivity between fragmented habitats. Effective
management of tropical seascapes must match the population biology and dispersal ability of
target species but the patterns of actual connectivity and the scale at which they occur remain
poorly understood. Identifying fish movements during their lifespan will contribute to a better
understanding of actual connectivity in the tropical seascapes and is a key-objective in
ecosystem conservation planning, by improving the knowledge required for the design of
MPA networks (Almany et al. 2009, Sale et al. 2010).
Revealing movement patterns and uses of multiple habitats during entire fish life requires
empirical studies of both larval and post larval dispersal, the latter consisting in juvenile and
adult mobility patterns. Fish movements and habitat use can be estimated through different
methods involving natural or artificial markers (Gillanders 2009, McMahon et al. 2011).
Among methods of direct measurement of connectivity, microchemistry of otoliths is the only
method which permits to identify and track back movements of fish during their entire life
(Cowen et al. 2007, Elsdon et al. 2008).
Otoliths consist in calcareous structures located in the inner ear of fish. They continuously
grow with successive deposits of calcium carbonate (CaCO3) mainly originating from the
surrounding water masses. During this calcification process, trace elements from the
environment are incorporated and stored in their carbonate matrix. Therefore, otoliths
constitute a natural recorder of waters masses occupied by fish during their life.
Environmental history may then be revealed by the microchemistry signatures recorded along
otolith growth, which in some case allows to reconstruct migratory paths of fish (Campana
1999, Elsdon & Gillanders 2003a).
138
Study of life history movements through microchemistry of otoliths requires chemical
heterogeneity of water masses in order to chemically distinguish specific signatures of
habitats or sites (Elsdon et al. 2008). That is why studies using microchemistry first focused
on diadromous species (Secor 1992, Secor et al. 2001, McCulloch et al. 2005), and then on
temperate marine species using distinct habitats during their ontogeny (Gillanders 2002,
Brown 2006b, Shima & Swearer 2009). Studies that have tried to reconstruct the life history
movements of coral reef fish using otolith microchemistry generally reported low rate or even
sometimes an absence of chemical discrimination between reef habitats (Chittaro et al. 2006,
Berumen et al. 2010), suggesting the technique may not be useful in coral reefs.
Lutjanus fulviflamma is a common Indo-Pacific snapper which uses mangroves as nursery
habitat and coral reefs as adult habitat (Thollot 1992b, Igulu et al. 2011, Berkstrom et al.
2013). This spatial segregation between adult and juvenile populations may imply distinct
chemical signatures between life stages. Within the adult life stage, fish may use several types
of reef, from the most coastal fringing reefs to the most oceanic barrier reefs, and perhaps
some temporary use of mangroves as well. The identification of movements among reef types
using microchemistry will depend on the chemical heterogeneity of reef habitats, and may
thus be possible only in a few place of the world such as New-Caledonia, a continental island
of the South-Pacific which possesses one of the world greatest nickel deposits and is
characterized by a highly complex and heterogeneous geochemistry (Paris 1981, Ambatsian et
al. 1997, Baille et al. 2012).
The aim of this study is to explore the patterns of movements and therefore, the use of
multiple habitats during the lifespan, of L. fulviflamma within the tropical seascapes of New
Caledonia. More precisely, we used otolith microchemistry to investigate if all juveniles of L.
fulviflamma stayed in mangrove during a similar length of time or if the patterns varied
among individuals or are site-specific. Then, we investigated if the transition from juvenile to
adult habitat was direct to a single reef habitat or gradual with intermediate reef habitats.
Finally, we studied whether adults moved permanently to reefs or if they returned at time in
mangroves.
139
2. Materials and methods
New Caledonia is a French archipelago located in the South West Pacific, 1500 km east of
Australia (Fig. 6.1). The archipelago is composed of a main island (Grande Terre) and several
groups of smaller islands. The main island held the largest lagoon of the world, with a linear
barrier reef of 1744 km long and a surface area of 31 336 km2 (Andréfouët et al. 2009).
Distance between the main island and the barrier reef varies from 1 to 65 km (Paris 1981) and
some parts of the lagoon are on the Unesco heritage list since 2008.
sites: Gatope (1), Ouano (2), St Vincent (3) and Prony (4).
The blackspot snapper Lutjanus fulviflamma (Forsskål 1775, Lutjanidae) is a small snapper
commonly targeted by subsistence fisheries throughout the Indo-Pacific. It is a carnivorous
species with a maximum total length (TL) of about 35 cm, and a mean size of 25 cm (TL).
This species has a complex life cycle and can be found in different habitats according to its
ontogenetic stage. Adults inhabit coral reefs whereas juveniles use shallow-waters habitats,
principally mangrove (Loubens 1980, Thollot 1992a, Carpenter & Niem 2001a, Igulu et al.
2011, Berkstrom et al. 2013). Because the movement of the species may partly blur the
characterization of chemical signatures in otoliths, we used the highly sedentary damselfish
Dascyllus aruanus (Cuvier 1829, Pomacentridae) as a control species. This small (maximum
140
TL of about 7 cm) damselfish lives in Acropora coral heads (Sale 1971) and its restricted
home range combined with its strong site fidelity from settlement to adulthood may result in
otoliths chemical signatures reflecting more accurately the chemistry of ambient water
(Chittaro & Hogan 2013).
Fish were sampled for otolith analysis in different habitats at 4 sites around the main land of
New Caledonia during three years (2009, 2010 and 2011) (Fig. 6.2). L. fulviflamma were
collected in reef habitats by spear fishing and juveniles were collected in mangroves using
gillnets with small meshes (7 mm) surrounding an isolated mangrove tree. Gillnets were
deployed at high tide and fish collected with handnets in a few cm of water and/or on the
ground at low tide. D. aruanus were anesthetized with eugenol and collected with handnets.
Immediately after collection, fishes were stocked in a freezer (at -20°C) until dissection.
The first sampling took place during austral summer 2009 in the mangroves and inner barrier
reefs of two sites: Gatope and St Vincent (Fig. 6.2A, 6.2C and Tab. 6.1). The second and third
samplings have been carried out during austral winter 2010 and 2011 in the mangroves,
fringing, intermediate and inner barrier reefs at four sites: Prony, St Vincent, Ouano and
Gatope (Fig. 6.2). With fish collected in different years, seasons and habitats, our sampling
included possible spatio-temporal variability of chemical signatures, which is essential to
characterize robust chemical signatures in otolith (Mercier et al. 2012).
In order to characterize the environmental signatures of mangrove, fringing reef, intermediate
reef and barrier reef habitats, up to five fishes from each sample were randomly selected for
otolith analysis, with a total of 284 otoliths analyzed (Tab. 6.1A). Another five fishes were
selected in the 2010 samples of the barrier reef habitat of each site in order to reconstruct fish
environmental history over the entire life of 20 adults L. fulviflamma (Tab. 6.1B).
141
Figure 6.2: Details of habitats sampled in St Vincent (A) Ouano (B), Gatope (C) and Prony
(D).
Table 6.1: Sample size to (A) characterize habitats chemical signatures by LA-ICPMS of the
surface of otoliths and (B) determine the patterns of fish movements by LA-ICPMS of
transverse sections of otolith along transects.
Objectives
Year
Site
Habitat
Laser analysis
Specie
n otoliths
A - Characterization of habitat
signatures
2009
2
M, BR
Ablation on surface
L.fulviflamma
D. aruanus
20
20
2010
2011
4
M, FR, IR, BR
Ablation on surface
L.fulviflamma
D. aruanus
125
119
2010
4
BR
Transects
L.fulviflamma
20
B - Patterns of fish movements
142
2.2. Otolith preparation
All material used for otolith extraction or preparation was decontaminated in 5% ultrapure
nitric acid bath (24h), rinsed three times with ultrapure water (18.2 MΩ), dried and stored in
plastic bag under a laminar flow hood (HEPA 100).
At the lab, fish samples were unfrozen and each individual was measured (Fork length, FL) to
the nearest mm and weighted to the nearest g. Paired sagittal otoliths were extracted using
ceramic tweezers, cleaned with ultrapure water, dried under a laminar flow hood (HEPA 100)
and stored in individual plastic vials.
To characterize environmental signatures, otolith external surfaces of fish collected in various
habitats (Tab. 6.1A) were analyzed with a Laser Ablation Inductively Coupled Plasma Mass
Spectrometry (LA-ICP-MS). Prior to analyses, otoliths were cleaned to remove any remaining
tissue following the Warner et al (2005) method with a first bath of 50/50 H2O2 (30%,
Suprapur®) and NaOH (0.1 mol.L-1, Suprapur®) solution during 1 hour. They were sonicated
during the last 5 min, then rinsed 5 times with ultrapure water for 5 min, dried under a laminar
flow hood (HEPA 100) and stored in individual plastic vials until LA-ICP-MS analyses.
To predict the environmental life history and reconstruct the movements of 20 adults L.
fulviflamma (Tab. 6.1B) transverse sections of otoliths were analysed by LA-ICP-MS along
transects from core to edge. Right sagittal otoliths were embedded in resin (Araldite 2020®)
and transverse sections including the otolith core were realized with a low speed saw
(Buehler®). Sections were then polished using decreasing sandpaper (from 800 to 1600
grains.cm-2), lapping film (9 to 1 µm) and diamond-tipped powders (Struers®, 1 to 0,3 μm),
rinsed in ultrapure water, dried under laminar flow hood (HEPA 100), mounted on a
microscope slide (10 sections per slide) with double-side tape and stored in plastic boxes until
LA-ICP-MS analyses.
2.3. Otoliths chemical analyses
Otoliths of 2009 and 2010 samples were analyzed by LA-ICP-MS at the Pôle de
Spectrométrie Océan (Institut Universitaire Européen de la Mer, Brest, France) using a
Thermo Element 2 coupled to a laser 193 nm CopexPro 102 Coherent. Otoliths sampled in
143
2011 were analyzed at the Resources Institute of Environmental Livelihoods (Charles Darwin
University, Darwin, Australia) using an Agilent 7500ce coupled to a UP – 213 nm laser
ablation system. The following 21 isotopes were measured for habitat discrimination: 7Li, 11B,
85
Rb, 88Sr, 95Mo, 111Cd, 117Sn, 138Ba, 208Pb, 232Th, 238U, 25Mg, 43Ca, 44Ca, 47Ti, 51V, 52Cr, 55Mn,
60
Ni, 65Cu and 66Zn. Both ICP-MS were operated at low resolution using argon as the carrier
gas. The laser was set at 5 Hz frequency and 90 µm beam diameter. Each ICP-MS analysis
consisted in 30 s background followed by 90 s ablation (laser activated).
To reveal the latest elements incorporated, and thus the chemistry corresponding to the habitat
of collection, a single laser ablation was done on the external surface of selected otoliths (Tab.
6.1A). To standardize the analyses, the ablation was always done at the same spot at the tip of
the post rostrum on the otolith posterior side.
To study patterns of movements during the entire life of each selected adult snapper (Tab.
6.1B), chemical transects were analyzed on fish otolith transverse sections. A transect
consisted in successive laser ablations in point-by-point mode from the core to the edge of the
otolith by increment of 140 µm and along the longest otolith growth axis.
To compensate for possible variation in quantity of material ablated, calcium was used as an
internal standard. An external standard (National Institute of Standards and Technology,
NIST, 612) was analyzed to correct instrument drift every ten ablations for transects and 10
otoliths for surface analyses.
Raw ICPMS data (counts per second, CPS) were transformed in elemental concentrations
(parts per million, ppm) following Longerich et al. (1996) method. Any value greater than
three times the inter-quartile distance were defined as outlier data and removed of data set
(Tukey 1977). Limits of detection (LOD) were also calculated following Longerich et al.
(1996) method and concentrations of elements inferior to the LOD were set to zero. Trace
element selected for statistical analyses had to follow two further criteria 1) elemental
concentrations had to be greater than the LOD in 70% of the otoliths in at least one particular
habitat (e.g. barrier reef) or one particular site (e.g. Gatope) and 2) the coefficient of variation
of elemental concentration in the NIST 612 had to be less than 10% (Chittaro et al. 2004,
Chittaro et al. 2006).
144
Even if LA-ICP-MS parameters were identical, the use of two different instruments may
induce variation in chemical analyses. To address this issue, all elemental concentrations were
standardized in percent of all measured elements.
2.4. Age and growth back-calculations
In order to correlate fish movements with age and size, snappers were aged by counting the
number of yearly increments (opaque bands) on otoliths and size-at-age were back-calculated
from measurement of otolith radius at age using the Dahl-Lea back-calculation model (Lea
1910):
where Ri and Li are otolith radius and fish length at age i, Rcpt and Lcpt are otolith radius and
fish length at capture. We assumed that one annulus is formed each year as it is the case for
several Lutjanidae species closed to study species (Newman et al. 1996, Newman & Dunk
2002).
2.5. Data analyses
All statistical analyses were performed using the statistical R software (R development Core
Team, 2011). To test the effect of habitat on the multi-elemental composition of otoliths,
multivariate one-way PERMANOVA based on Euclidian distance were performed (Legendre
& Anderson 1999). To compare variation in chemical concentrations of each element between
habitats, univariate one-way PERMANOVA test were performed. For both tests, observations
were permuted 999 times. When differences were significant, multiple comparisons among
mean ranks were performed in order to identify which habitats were significantly different
and for which element (Siegel & Castellan 1988). Non-parametric methods were used since
the assumptions of normality and homogeneity of variance required by parametric
MANOVAs were not met.
Random Forests (RF) algorithm (Breiman 2001) was performed on elemental composition of
otoliths surface to build an habitat classifier based on otolith microchemistry. RF is a machine
learning classification method allowing freedom from normality and homoscedasticity
145
assumptions. RF is also the most accurate method for classification of otolith chemical
signature (Mercier et al. 2011, Mercier et al. 2012). With the RF method, a random
subsample of about two-third of the dataset is used to build a classification tree and the
remaining one-third is used to estimate classification accuracy of this tree. This process is
reiterated many times to build a random forest of classification trees, 5000 trees in our case, in
order to ensure that every individual gets predicted several times, which permits to estimate
classification accuracy of the method (percent of correct classification). Here, all possible
combinations of elements were tested by RF in order to find the combination that showed the
highest percent of correct classification (Mercier et al. 2011, Mercier et al. 2012). In case of
multiple best combinations, the one composed of the smaller number of trace elements was
retained.
A PCA was also used to further visualize the elemental composition of otoliths from fish
collected in different habitats. Elemental compositions were arcsin√x transformed in the PCA
to normalize the data.
Paillon et al. (In prep.) (see Chapter 3) indicated strong differences in otolith chemistry
between mangroves and reefs but possible confusions among reef types for L. fulviflamma in
New Caledonia. They suggested that the latter may possibly be due to the intermediate
position of patch reefs in the lagoon which, depending on sites, may be more influenced by
coastal waters and therefore chemically closer to fringing reefs, or more influenced by
oceanic waters and thus chemically closer to barrier reefs. Consequently, we adopted here a
two-step strategy to predict habitat from otolith microchemistry. First, a RF habitat classifier
was produced to distinguish mangroves from reefs using the chemical analyses of L.
fulviflamma otoliths of all sites. Second, for each site, a RF classifier was built to distinguish
inshore vs. offshore reefs, by either pooling fringing with intermediate reefs or pooling
intermediate with barrier reefs. This procedure was based on the otoliths elemental
compositions of the control species D.aruanus since this species is very sedentary and more
prone to represent environmental conditions. The pooling that produced the highest percent of
correct classification from the D.aruanus dataset was used to build RF classifiers for the same
inshore and offshore pooling of reefs for each site with the L. fulviflamma dataset.
146
The two RF classifiers obtained for L. fulviflamma were then successively used on otolith
transects to predict lifespan habitat use and movement of the species. First, each point of each
transects were predicted either as mangrove or reefs using the first RF classifier. Then, the
second RF classifier was used to predict whether reef predictions were inshore of offshore
reefs. Because habitat predictions were obtained from a forest of 5000 trees, each tree
providing a vote for a habitat, the analysis not only provided the most probable habitat (i.e.
the habitat gathering the majority of the 5000 votes) but also the actual probability of the
prediction (i.e. the % of votes).
3. Results
3.1. Habitat classification
Twelve elements responded positively to the criteria of selection and were selected for
statistical analyses: B, Ba, Cr, Mg, Mn, Pb, Rb, Sn, Sr, Th, U and Zn. Multivariate
PERMANOVA showed that elemental composition of L. fulviflamma otoliths differed
significantly between mangrove and reefs (F
(1,143)
= 12.83; p = 0.003). Univariate
PERMANOVA revealed significant differences between habitats for five elements: Mn, Rb,
Sn, Sr and U (Fig. 6.3). Mn and Sn were in greater concentrations in otoliths of individuals
collected in mangrove. Rb, Sr and U were in greater concentrations in otoliths from reefs.
Figure 6.3: Otolith mean value of Mn:Ca, Rb:Ca, Sn:Ca Sr:Ca and U:Ca ratios per habitat
(mangroves and reefs). Mean and standard error are plotted. Statistical significances of oneway anova results are indicated above graphs (“***” p-value < 0.001;; “**” p-value < 0.01;
“*” p-value < 0.05;; “+” p-value < 0.1).
147
The best RF classifier of reefs and mangroves only used 3 elements: Mn, Pb, Rb (Tab. 6.2).
With this combination, 92% of individuals were correctly classified. Percent of correct
classification were 85% for fish collected in mangroves and 98% for fish collected in reefs.
Mn and Rb were the strongest characteristics of mangroves and reefs, respectively (Fig. 6.4).
Figure 6.4: PCA biplot of individual fish collected in (M) mangroves and (R) all reefs. Only
concentrations of Rb, Pb and Mn in otoliths were used in the PCA.
For Gatope, Ouano, and Prony, data on D. aruanus otoliths showed that RF classifications of
reefs were best when intermediate and barrier reefs were pooled in an offshore reefs group.
On the opposite, classification of reefs at St Vincent was most accurate when fringing and
intermediate reefs were pooled in an inshore reefs group. The percentage of correct
classification for fish collected in inshore reefs varied from 78% to 100% depending on the
site (Tab. 6.2). For offshore reefs, the percentage of correct classification varied between 89%
and 100%. Overall, between 90% (Prony) and 95 % (Gatope and St Vincent) of individuals
were classified into the correct reef category for D. aruanus (Tab. 6.2).
Using the same reef grouping, the percentage of correct classification for L. fulviflamma
varied from 70% at Ouano to 94% at Gatope (Tab. 6.2). Fish collected in inshore reefs
showed a minimum of 50% correct classification at Ouano and a maximum of 80% at Gatope
148
and Prony. Percentage of correct classification for fish from offshore reefs ranged from 72%
at St Vincent to 100% at Gatope.
Table 6.2: Random Forest classifications with details of the best combinations of elements
and the percentage of correct classification for each analysis (M: mangrove; FR: fringing reef;
IR: intermediate reef; BR: barrier reef).
Optimal combination Correct classification (%)
Specie
Site
L. fulviflamma All (4)
Discrimination
Elements
Individual
Mangrove
Reefs
M vs Reefs
Mn, Pb, Rb
92
85
98
Individual
Inshore
Offshore
Gatope
FR vs IR/BR
Ba, Cr, Mg, Sr
95
78
100
Ouano
FR vs IR/BR
B, Pb, Sn, Th
93
100
89
Prony
FR vs IR/BR
Ba, Mg, Sr, U
90
80
95
StVincent
FR/IR vs BR
B, Ba, Cr, Pb, Sn, U
95
100
90
L. fulviflamma Gatope
FR vs IR/BR
Cr, Mn, Sn
94
80
100
Ouano
FR vs IR/BR
Sn U
70
50
80
Prony
FR vs IR/BR
Ba Mg Mn Sr
85
80
88
StVincent
FR/IR vs BR
Cr Mg Pb Rb Sr Zn
75
78
72
D. aruanus
3.2. Predictions of environmental life history
Reconstruction of microchemistry of entire life provided a high level of global accuracy.
Nearly half of habitat predictions (42% of total habitat predictions) had a probability ranging
from 75% to 100% of the votes. More than half of the predictions showed probabilities
comprised between 50% and 75% (56% of total habitat predictions). The rest of the
predictions had a probability ranging from 25% to 50% of the votes (2% of total habitat
predictions). For fish collected in Gatope and Prony, the probabilities of RF habitat
predictions were >75% for more than half of habitats predicted, the rest of the predictions
were comprised between 50% and 75%. Ouano and St Vincent predictions showed inferior
percentage of votes, with a majority of predictions comprised between 50% and 75% and less
than half of the predictions >75% probabilities. In all sites, the probabilities of RF habitat
predictions comprised between 25% and 50% constituted 2% of the total predictions realized
which was considered as negligible.
The reconstruction of environmental life history revealed one main life history trait. All
individuals had a juvenile phase in mangrove and an adult phase on reefs (see Appendix 12).
149
Three different patterns of movement were identified depending on the fish (Fig. 6.5). The
first pattern showed four individuals evolving in two different habitats, mangrove as juveniles
then offshore reef as adult. Two of these fish were collected in Gatope and two in Ouano (Fig.
6.5A). The second pattern showed fish evolving in three different habitats: mangrove as
young, then inshore and offshore reefs as adults with no return to mangrove for adults but
random movements between inshore and offshore reefs (Fig. 6.5B). This pattern was observed
for five fishes: one from Gatope, two from Prony and two from St Vincent. The last pattern
consisted of random movements between the three habitats (Fig. 6.5C). It was observed for
eleven individuals, and was therefore the most common pattern. However, there were no
obvious relationship between the site and type of patterns as an approximately equal number
of fish displayed this pattern at each site (two at Gatope and three at each of the three other
sites).
Fish ranged from 4 to 18 years in age and from 18 to 28 cm in fork length. Six individuals left
the mangrove before the end of their first year with a mean size of 4 cm. Interestingly, all
these six fish showed exclusively the third pattern of movements with some return in
mangrove after first departure. Nine fish left the mangrove during their second year with a
mean size of 14 cm. Among them, one individual displayed the first pattern of movements,
four displayed the second and four displayed the third. Five fish left the mangrove during
their third year with a mean size of 17 cm. Among them, three individuals displayed the first
pattern of movements, one displayed the second and one displayed the third. There was no
site-specific trend between the time spent in mangrove and the age or body length of fish.
150
Figure 6.5: Representation of the three main patterns of lifespan movement for L.
fulviflamma using otoliths microchemistry and RF predictions. The x axis represents the
otolith radius and the y axis the three possible habitats occupied during life. The second top x
axis indicates the age and length of the fish (respectively below and above axis). The grey
gradient shows the probabilities of habitat predictions (% of vote). The white line joins
habitats predicted with the highest probability at each LA-ICPMS analysis.
151
4. Discussion
Several studies conducted in the Caribbean and the Indian Ocean have shown that many coral
reef fish use mangroves as juveniles (Nagelkerken et al. 2002, Cocheret de la Morinière et al.
2003a, Chittaro et al. 2004, Dorenbosch et al. 2004, Dorenbosch et al. 2006). However, the
pattern of use of mangrove in the South Pacific region is largely unknown (Thollot 1992a,
Sheaves 2005, Unsworth et al. 2008). The high levels of mangrove and reef discrimination
obtained with RF during this work permitted to study several characteristics of the blackspot
snapper environmental life history and to understand the functional connectivity among reefs
and mangrove in a continental island of the South Pacific, New Caledonia. First, we estimated
the use of mangrove as juvenile habitats for this coral reef species as well as the patterns of
transition between juvenile and adult habitats. Then, we investigated the different patterns of
movements between mangroves and several reef categories showed by adults over their entire
life.
4.1. Classification accuracy and the grouping of reefs
Inshore and offshore reef groups formed using D. aruanus otolith datasets may partly be
explained by environmental characteristic of each site. For Gatope, Ouano, and Prony,
classification accuracies of reefs were better when intermediate and inner barrier reefs were
pooled in the offshore reefs group. On the opposite, St Vincent classification accuracies were
better when fringing and intermediate reefs were pooled in the inshore reefs group. Even if
Gatope intermediate reef was spatially closest to the fringing reef, it was environmentally
closest to the inner barrier reef because of the proximity of a deep channel connecting the
lagoon to the open ocean. For Ouano, the distinction between the three types of reefs was less
obvious because they were all spatially close to one another with, in addition, a fringing reef
probably under the influence of a channel bringing oceanic waters. Nevertheless, the
intermediate reef was closest to the barrier reef than the fringing reef, suggesting that in
Ouano, intermediate reefs are more influenced by oceanic waters than coastal ones. Prony is
an enclosed bay with a fringing reef located deep inside the bay with strong influence of
freshwater inputs and an intermediate reef located at the exit of the bay, equally distant to the
fringing and barrier reefs, and under the influence of the same channel than the inner barrier
reef. This particular spatial distribution of reefs in Prony may partly explain the
environmental proximity of intermediate and barrier reefs at this site. Finally, in St Vincent,
152
fringing reefs are located along the shore of a series of high islands which extend inside the
lagoon and are progressively replaced by coralline islands surrounded by intermediate reefs,
thus making the environmental proximity of fringing and intermediate reefs logical in this site
(Fig. 6.2).
The grouping of reefs in inshore and offshore categories from datasets collected in the otoliths
of D. aruanus, a small and highly sedentary species, worked relatively well for L.
fulviflamma, a medium-sized relatively mobile species. For only one site, Ouano, inshore reef
classification accuracy was high for D. aruanus (80%) and low for L. fulviflamma (50%). This
difference was also observed, but to a lesser extent, for the offshore reefs of this site (89
versus 80% correct classification), and may be explained by the spatial proximity of the reefs
and the probable frequent movements among these reefs for the mobile species. This pattern
was confirmed by a recent study of L. fulviflamma movements using acoustic telemetry at
Ouano (Chateau et al. 2012).
4.2. Pattern of juvenile habitat use and transition between juvenile and adult
populations
As shown by many studies, mangroves are known to be a juvenile habitat with more or less
importance depending on the geographical area and the species considered (Parrish 1989,
Beck et al. 2001, Dorenbosch et al. 2005a, Adams et al. 2006). Mangroves provide shelters
and abundance in small size prey for zoobenthivores juvenile snappers. In the Indo Pacific,
mangroves are known to be an essential habitat for L. fulviflamma juveniles (Thollot 1992a,
Igulu et al. 2011, Olds et al. 2012, Berkstrom et al. 2013) and in New Caledonia they may
even be obligatory (see Chapter 5). Except for one fish which stayed less than one year in
mangroves as juvenile, all fish in this study stayed in mangrove during at least their first year,
and for many individuals their first 2-3 years. Then, juveniles or sub-adults left the
mangroves, definitely or for a variable length of time, to reach coral reefs before the end of
their third year.
L. fulviflamma movements between mangrove and coral reefs may be associated with an
ontogenetic shift in diet (Berkstrom et al. 2013). In Tanzania, even if the main eaten preys
were brachyurans for all size classes of L. fulviflamma, there was a gradual ontogenetic shift
153
in diet for the secondary prey groups, mostly fish. Indeed, juvenile and sub-adult diets were
mainly constituted of small shrimps and prawns which were replaced by demersal teleosts, the
most abundant prey of the adult blackspot snapper (Cocheret de la Morinière et al. 2003b,
Kamukuru & Mgaya 2004).
There was no site-specific trend for time spent in mangrove, or age or length of fish when
they left their juvenile habitat. Nevertheless, all juveniles which left mangrove before their
first year showed the third pattern of movement which involved visits to mangrove as adults.
However, the transition between coastal juvenile habitats and adult reef habitats was generally
direct toward offshore reefs. A few of the individuals showed a gradual transition, passing
briefly through inshore reefs before reaching offshore reefs. This observation may be a real
pattern or simply reflects an inshore reefs transition that was too brief to be chemically
marked in otoliths, and/or an inshore reefs signature that was not distinct enough to be
chemically discriminated from the offshore signature.
4.3. Life time movement patterns of the blackspot snapper
Reconstruction of environmental life history of the blackspot snapper revealed intra-specific
variation in three different patterns of movement. The simpler one involved individuals using
only two habitats during their life, mangrove as juvenile habitat and offshore reefs as adult
habitat. The second pattern was more complex with the use of three habitats, only mangrove
as juvenile habitat and both inshore and offshore reefs as adult habitats. In this pattern, some
movements occurred between inshore and offshore reefs but there was no return to mangrove
during the adult stage of the snapper. Finally, the third pattern of movement was the most
common as it was observed in over half of the studied fish. This pattern consisted in multiple
movements between reefs with short or long term returns to mangrove.
The intra-specific variation in patterns of movements was confirmed for L. fulviflamma in
Ouano by a study using acoustic telemetry as artificial tag (Chateau et al. 2012). This study
showed that L. fulviflamma presented multiple movements between mangroves and reefs
(nearby reefs or distant reefs) as adults and that these movements’ extent and duration varied
according to individuals.
154
In their study, Hammerschlag-Peyer and Layman (2010) indicated that intra-specific
movements and habitat use could significantly vary among individuals. These differences
may be attributed to a polymorphism in behavioral traits, some phenotypic plasticity (bold or
shy behavior), and/or different responses of individuals to competition for food and shelters.
In New Caledonia, the segregation between juvenile and adult habitats of the black-spot
snapper was partial since some adult returned to mangrove. On tropical coasts, mangroves are
used as foraging grounds by many fish from adjacent habitats because they provide trophic
resources. This trophic role does not concern exclusively juveniles but also involves adults
(Sheaves 2005). Mangroves may be an obligatory nursery habitat for juveniles of L.
fulviflamma but they may be also a feeding habitat for sub-adults and adults. A gut content
and stable isotopic analysis of L. fulviflamma in New Caledonia may provide details on
activity of adult blackspot snappers visiting mangroves.
4.4. Limits of the method
Movements of adults and juveniles may occur at the local scale, between habitats within site
(less than 10 km), to regional scale, between sites (more than 50 km) (Grüss et al. 2011).
However, as Chittaro study (2005), environmental chemistry of New Caledonia tropical
seascape was not distinct enough to identify site-specific chemical signatures for each habitat
(ex. mangrove of Ouano). Estimate of movements between habitats have therefore been
realized by taking each site independently and could not been studied between sites. This
issue may not be a problem for our study as Kaunda-Arara and Rose (2004b) showed in their
tracking study that L. fulviflamma could travel distances around 2 km corresponding to
movements between patch habitats within a site and not among sites.
Furthermore, reconstruction of movement pathways with microchemistry could not provide a
high temporal resolution due to the size of the laser ablation. We had to face a compromise
between temporal resolution and the amount of material ablated. Using an extremely small
laser ablation would have provided greater temporal resolution but will not have permitted the
measurement of some important trace elements for habitat discrimination. In this context,
daily movements linked to foraging activity could not be estimated with microchemistry.
Other methods such as acoustic telemetry are complementary to otolith chemistry and can
155
provide information on connectivity at smaller spatial and temporal scales (Hitt et al. 2011,
Chateau et al. 2012).
5. Conclusion
Functional links among habitats constituting a tropical seascape were mainly studied in the
Caribbean region (Nagelkerken et al. 2000, Nagelkerken et al. 2002, Chittaro et al. 2004,
Mateo et al. 2010), rarely in the Indo-Pacific region (Dorenbosch et al. 2005a, Nakamura et
al. 2008, Unsworth et al. 2008, Unsworth et al. 2009, Berkstrom et al. 2013) and even less in
the South Pacific (Thollot & Kulbicki 1988, Dorenbosch et al. 2005a, Sheaves 2005,
Berkstrom et al. 2013). In this context, it has been suggested that coastal shallow-waters
habitats such as mangrove and seagrass beds were less connected with coral reefs in the IndoWest Pacific region than in the Western Atlantic (Parrish 1989, Sheaves 2005). Besides,
Thollot (1992a) showed that few interactions existed between mangroves and coral reefs in
New Caledonia. The aim of this work was to demonstrate that biological connectivity
between mangroves and coral reefs was effective for one species in New Caledonia. Indeed,
mangroves and coral reefs of New Caledonia were interconnected for the blackspot snapper
by ontogenetic migrations and by adult movements between the two ecosystems.
This work contributes to knowledge building on movement pathways and use of multihabitats by coral reef fish in tropical seascapes. Collecting empirical data with direct
measurements of movements is essential to improve fundamental knowledge of biological
connectivity. This knowledge may provide powerful tools for spatial management planning,
especially in the design of MPA networks which include coastal shallow-waters habitats such
as mangrove as well as coral reefs in order to protect all habitats required by organisms to
complete their life cycle. At present, most reserves fail to incorporate connectivity into their
management frameworks (Almany 2009). For fish, the scales at which actual connectivity
takes place at the different life stages are largely unknown (Di Franco et al. 2012).
156
Caledonia.
University of New Caledonia, the Pole of Ocean Spectrometry of the European University
Institute of the Sea and the Research Institute of Environment and Livelihoods of Charles
Darwin University. We specifically acknowledge Gerard Mou-Tham, Joseph Baly, and
Miguel Clarque for their invaluable field and laboratory assistance, Claire Bassoulet and
Françoise Foti for their precious help with the LA-ICP-MS analyses.
157
Le chapitre 6 a permis d’estimer la connectivité réelle entre les habitats du lagon de
Nouvelle-Calédonie au cours de la vie de Lutjanus fulviflamma. Cette étude identifie une
variabilité intra-spécifique des mouvements entre les habitats puisque trois patterns de
mouvements aux échelles spatiales ascendantes ont pu être révélés. Un pattern à petite échelle
spatiale où, une fois la mangrove quittée, les individus adultes évoluent au sein d’un habitat récifal unique. Un pattern à échelle spatiale intermédiaire où les individus se cantonnent à des
mouvements restreints entre les différents habitats récifaux. Un pattern à échelle spatiale
étendue où les individus montrent des déplacements de l’ordre de plusieurs kilomètres, alternant entre récifs barrières et retour vers la mangrove.
Par ailleurs, cette étude a aussi permis de préciser l’importance de la mangrove au cours
de la vie de L. fulviflamma. Outre l’utilisation de la mangrove comme habitat juvénile obligatoire (Chapitre 5), la majorité des individus adultes effectuent des mouvements entre les
récifs et la mangrove, ce qui indique le caractère essentiel des mangroves pour les adultes de
cette espèce. Une étude parallèle par marquage acoustique suggère que les adultes de L.
fulviflamma visitent les mangroves pour la recherche de nourriture (Chateau et al. 2012).
Les patterns de mouvements peuvent être influencés par la configuration du paysage
marin constitué d’habitats continus ou fragmentés. Les habitats continus facilitent les
mouvements sur de longues distances alors que les habitats fragmentés ont tendance à
entraver les mouvements. Cependant, aucune tendance de mouvement spécifique aux sites et
à leur géomorphologie n’a pu être observée au cours de notre étude. L. fulviflamma est une
espèce qui de par sa taille, son home range et sa capacité à se déplacer sur de longues
distances est susceptible de pouvoir franchir les barrières physiques d’un paysage fragmenté (Kramer & Chapman 1999, Kaunda-Arara & Rose 2004b, Sale et al. 2005). L’estimation de la mobilité des juvéniles et adultes et l’identification des habitats utilisés au cours de leur vie est d’une importance significative pour le fonctionnement des aires marines protégées car elle
souligne la nécessité d’inclure de multiples habitats au cours de leur design (Grüss et al.
2011).
158
159
160
Chapitre 7 : Discussion générale et perspectives
161
162
Chapitre 7 : Discussion et perspectives
Chapitre 7 : Discussion générale et perspectives
Les paysages marins tropicaux sont constitués de multiples habitats fragmentés qui
abritent des populations locales reliées par des échanges d’individus. La perte et l’érosion de ces habitats menacent la résilience des écosystèmes en diminuant la connectivité écologique
reliant les populations locales (Valiela et al. 2001, Hughes et al. 2003, Pandolfi et al. 2003,
Duke et al. 2007). La préservation de la connectivité écologique, nécessaire au maintien de la
résilience des écosystèmes, est donc un objectif clé pour la gestion efficace des écosystèmes
côtiers tropicaux. Pour cela, il faut pouvoir quantifier l’importance des liens fonctionnels entre les habitats et comprendre leur structure (Mumby 2008, Cowen 2006). Cependant, il est
remarquablement difficile de développer des estimations quantitatives de l’utilisation des habitats et des mouvements les connectant.
La microchimie des otolithes est l’outil le plus performant pour la mesure de la connectivité écologique au cours de la vie entière des poissons. Elle peut être particulièrement
utile pour l’identification des nurseries c'est-à-dire des habitats juvéniles contribuant
fortement au renouvellement des populations adultes, et pour l’estimation des habitats utilisés et des mouvements réalisés au cours de la vie des adultes. Avant d’utiliser la microchimie des
otolithes pour quantifier la connectivité écologique, il est néanmoins nécessaire d’approfondir les connaissances acquises sur cet outil, notamment en établissant son potentiel et ses limites
au sein des paysages marins tropicaux.
1. Axe méthodologique
Au cours de cette étude, le développement de l’outil microchimie des otolithes a été réalisé en plusieurs étapes visant à tester la fiabilité et le potentiel de cet outil en NouvelleCalédonie. Dans un premier temps, les variations spatiales de la microchimie des otolithes ont
été caractérisées et les signatures spécifiques des habitats élaborées pour plusieurs espèces de
poisson ainsi que pour l’environnement. Dans un second temps, l’utilisation de signatures multi-spécifiques comme proxy des signatures mono-spécifiques a été évalué. Enfin, dans un
troisième temps, les liens entre la microchimie des otolithes et de l’environnement ont été explorés.
163
L’hétérogénéité spatiale de la microchimie est le facteur prépondérant influençant l’efficacité de la discrimination spatiale et l’obtention de bonnes classifications. Les résultats de ce travail ont permis de révéler un potentiel élevé de l’outil microchimie des otolithes en Nouvelle-Calédonie avec une discrimination spatiale importante à l’échelle des habitats,
particulièrement des habitats fortement contrastés comme la mangrove et le récif barrière.
Cependant, l’efficacité de la discrimination spatiale diminue au niveau des stations, particulièrement dans le cas des récifs qui présentent une microchimie relativement homogène
entre eux.
Au cours du chapitre 3, il a été démontré qu’une variabilité interspécifique significative de la microchimie des otolithes existait au sein d’un même environnement. Par conséquent, l’utilisation de signatures multi-spécifiques en substitut de mono-spécifiques n’est en général pas possible même lorsque les signatures multi-spécifiques sont élaborées à partir de groupes
d’espèces taxonomiquement ou fonctionnellement proches. Cette variabilité interspécifique
pourrait résulter de traits de vie contrastés entre les espèces étudiées mais pourrait également
être liée à une variation interspécifique des processus d’incorporation des éléments au sein des otolithes (Gillanders & Kingsford 2003, Swearer et al. 2003, Barnes & Gillanders 2013).
Néanmoins, dans le cas d’habitats chimiquement très contrastés tels que les mangroves et
récifs barrières, les signatures multi-spécifiques apparaissent comme substituables aux
signatures mono-spécifiques. Ainsi, des similarités entre les espèces peuvent exister dans
l’incorporation des éléments de l’environnement vers les otolithes même si des différences interspécifiques semblent être le cas général. De ce fait, l’étude des relations entre la microchimie des otolithes et de l’environnement abordée au cours du chapitre 4 a permis d’approfondir les connaissances sur les interactions existantes entre les otolithes et l’environnement. Ce chapitre indique qu’il existe de fortes relations entre la chimie des otolithes, de l’eau, de la salinité et de la température. Néanmoins, ces relations sont complexes et dépendent des éléments et des espèces considérées. Les résultats démontrent que
la chimie des otolithes n’est pas l’exact reflet de la chimie de l’eau. L’influence spécifique de
chaque facteur, exogène et endogène, sur l’incorporation des éléments au sein des otolithes est extrêmement difficile à caractériser pour tous les éléments et espèces étudiés. Sans ces
connaissances, les interprétations écologiques issues de l’étude des signatures chimiques ne sont pas incontestables. Néanmoins, la forte structuration spatiale observée à partir de la
164
chimie de l’eau s’exprime au sein des otolithes, ce qui permet une utilisation cohérente de la
microchimie des otolithes comme traceurs des environnements passés.
Au vu de la perte engendrée par l’utilisation de signatures multi-spécifiques, le monospécifique reste donc le classifiant ultime et ne peut être substitué excepté dans de rares cas.
Ainsi, la reconstruction de l’histoire environnementale d’une espèce d’intérêt exige de caractériser les différents habitats à partir de la microchimie des otolithes de cette même
espèce. L’échantillonnage devient problématique lorsque l’espèce étudiée est rare ou emblématique bénéficiant d’un statut de protection (aspect conservation) ou simplement difficile à échantillonner (aspect logistique). Au contraire, si l’espèce est abondante sans statut de conservation particulier, l’utilisation de la microchimie des otolithes pour reconstruire
l’histoire environnementale des espèces permettra d’approfondir les connaissances sur leur biologie et d’estimer la connectivité réelle existante essentielle à la préservation des stocks d’adultes, notamment pour des espèces d’intérêt commercial. Cependant, il est important de
souligner que les signatures multi-spécifiques caractérisent très bien les habitats contrastés
tels que les mangroves et récifs barrières et peuvent donc être utilisées dans ce cas pour les
espèces protégées et/ou difficiles à échantillonner en nombre suffisant.
2. Axe écologique
Estimer la connectivité réelle en déterminant les mouvements au cours de la vie nécessite
de pouvoir tracer activement les mouvements d’individus entre les habitats ou d’identifier rétrospectivement les habitats fréquentés. La microchimie des otolithes utilise cette seconde
approche et permet d’identifier les nurseries, de déterminer les habitats utilisés au cours de la
vie des individus et les connexions existantes entre ces habitats.
Selon Beck (2001), un habitat est considéré comme une nurserie pour les juvéniles d’une espèce donnée si sa contribution à la production d’individus recrutant au sein des populations
adultes est supérieure à la production d’autres habitats où les juvéniles sont également présents.
Au cours de cette thèse, l’analyse microchimique de la partie juvénile des otolithes de Lutjanus fulviflamma a démontré une fréquentation systématique des mangroves durant la
phase juvénile. L’analyse de données de comptage et de la cartographie des mangroves a
165
montré que la distribution spatiale de l’espèce en Nouvelle-Calédonie est fonction de la
présence des mangroves. Cet habitat est donc clairement identifié comme un habitat juvénile
obligatoire pour L. fulviflamma en Nouvelle-Calédonie. La mangrove est assurément qualifiée
de nurserie pour cette espèce en Nouvelle-Calédonie car elle assure le lien vital vers les récifs
coralliens en transférant avec succès une biomasse de juvénile importante vers les populations
adultes (Beck et al. 2001, Dahlgren et al. 2006). Néanmoins, il est important de réaliser
qu’une certaine plasticité dans le choix des habitats juvéniles peut exister en fonction de la
zone géographique. Outre la mangrove, L. fulviflamma utilise aussi d’autres types d’habitats juvéniles tels que les herbiers dans l’Océan Indien (Igulu et al. 2011, Kimirei et al. 2011,
Berkstrom et al. 2013).
Les mouvements ontogéniques représentent des transferts de matière horizontaux et
unidirectionnels entre la mangrove et les récifs coralliens (Berkstrom et al. 2012). La
transition entre les deux écosystèmes peut être directe ou graduelle avec l’utilisation d’habitats intermédiaires identifiés comme les corridors spécifiques aux migrations
ontogéniques. Au cours de cette étude, un faible nombre d’individus a montré une transition
graduelle, ce qui suggère qu’une transition directe entre les habitats juvénile et adulte serait la règle pour L. fulviflamma. Cependant, il est aussi possible que la fréquentation d’habitats intermédiaires n’ait pu être identifiée en raison d’une fréquentation trop brève pour que les signatures microchimiques spécifiques des habitats intermédiaires soient marquées au sein de
l’otolithe. En effet, il a été démontré que l’incorporation de la signature microchimique d’un environnement au sein des otolithes requiert un laps de temps de 2 semaines en moyenne,
dépendant fortement du taux de croissance. Or, la microchimie des otolithes ne permet pas
toujours d’analyser les mouvements à cette échelle temporelle. Ainsi, la résolution temporelle
analysée va principalement dépendre du diamètre du faisceau laser de l’ICP-MS et du taux de
croissance de l’individu. Plus le diamètre du laser sera important, plus la surface ablatée correspondra à un laps de temps important de la vie du poisson. De même, plus l’individu sera vieux, plus sa croissance sera faible, et donc plus la surface ablatée correspondra à un laps de
temps important (Campana & Neilson 1985).
La connectivité est définie par les liens fonctionnels reliant les habitats et se traduit par
les mouvements des individus entre habitats et populations locales. Les résultats obtenus
durant cette thèse indiquent de remarquables mouvements pour L. fulviflamma avec trois
166
différents patterns de mouvements révélés. La multiplicité des patterns traduit une forte
variabilité intra-spécifique empêchant d’établir un unique modèle de mouvement pour cette
espèce.
Il n’y a pas de ségrégation spatiale stricte entre les populations juvéniles et adultes dans
l’utilisation des habitats puisque au cours de ces travaux, il a été mis en évidence que les individus utilisent la mangrove à la fois en tant que juvéniles mais aussi à l’âge adulte. Malgré son utilisation au cours des deux stades de vie, la mangrove garde toutefois son statut de
nurserie pour L. fulviflamma puisque les individus juvéniles ne sont pas présents au niveau
des habitats adultes. Les mouvements constatés à l’âge adulte entre la mangrove et les récifs
caractérisent des transferts de matière bidirectionnels entre ces deux écosystèmes.
3. Conclusion et perspectives de travail
La fragmentation du paysage marin tropical entraine une rupture potentielle des échanges
de gènes entre les habitats et tend à influencer la structure des métapopulations en augmentant
l’hétérogénéité génétique. Dans un contexte de fragmentation croissante, les mouvements
entre sous-unités sont donc cruciaux car ils permettent de connecter les populations locales et
d’augmenter les échanges génétiques.
Cette étude apporte des connaissances nouvelles sur la connectivité réelle existant dans le
lagon de Nouvelle-Calédonie qui sont essentielles à la gestion et la conservation des
écosystèmes. Par la mise en évidence directe de connectivité réelle entre les mangroves et les
récifs coralliens, ce travail de thèse souligne aussi l’importance du rôle des mangroves dans le Pacifique Sud. Cependant, ce travail a permis d’identifier l’importance des mangroves pour une unique espèce. Ces résultats encourageants incitent à de futures recherches ouvertes à
d’autres espèces récifales afin d’estimer l’importance globale du rôle de nurseries des
mangroves du Pacifique Sud. En effet, le rôle de nurseries des mangroves du Pacifique Sud
reste méconnu par rapport aux mangroves de l’Atlantique Est (Caraïbes) et l’Océan Indien (Afrique de l’Est). Si les mangroves du Pacifique Sud s’avéraient effectivement posséder un
rôle de nurserie essentielle pour de nombreuses espèces récifales, notamment pour des
espèces commerciales, l’inclusion des mangroves lors du design des réseaux d’aires marines protégées allant de la côte vers le large devrait être encouragé.
167
Dans l’optique de futures recherches, notamment si elles s’ouvrent à d’autres espèces que
L. fulviflamma, il serait intéressant d’ajouter l’étude des herbiers qui n’a pu être réalisée au cours de cette thèse pour des raisons logistiques. Les herbiers sont reconnus comme des
nurseries pour de nombreuses espèces (Mellin et al. 2007, Unsworth et al. 2008). De plus, les
herbiers sont souvent en bordure des mangroves et pourraient ainsi assurer un rôle important
au sein du continuum mangroves - récifs coralliens en tant qu’habitat corridor lors des migrations ontogéniques ou des mouvements des adultes entre ces écosystèmes.
L’utilisation de plusieurs ICP-MS (Montpellier, Brest, Darwin) au cours de cette étude a
été source de variations même si les réglages des différents appareils étaient identiques
(fréquence, diamètre, puissance du laser, etc.). De même, l’intervalle de temps trop élevé entre deux sessions d’analyse sur le même appareil (Brest, 2009 et 2011) s’est aussi révélé être problématique. Pour palier ces problèmes d’inter-calibration, les données ont été
standardisées et la plupart des analyses ont été réalisées non plus en concentrations absolues
(en ppm) mais en concentrations relatives (en pourcentage de chaque élément mesuré). Cette
transformation engendre une perte d’information non négligeable et dans l’optique de futures recherches, il serait préférable de travailler à l’aide d’un seul appareil et d’éviter un délai trop important entre deux sessions d’analyses. La résolution temporelle analysée au cours des transects pourrait être affinée en
diminuant le diamètre du faisceau laser mais au prix d’un temps d’analyse plus important et d’une baisse de la sensibilité de détection des concentrations. En effet, réduire le diamètre du faisceau afin d’augmenter la résolution temporelle des variations de concentration se fera au
détriment de la quantité de matériel ablaté et donc du nombre d’élément supérieur aux limites de détection. Dans l’optique de futures recherches, l’obtention d’une résolution temporelle plus fine pour un temps d’analyse acceptable pourrait être abordée via la comparaison de deux
méthodes d’analyse : le transect point par point (fixe) et le transect continu (à vitesse
constante).
Dans le but d’augmenter la discrimination des habitats, il serait intéressant d’inclure lors
de futures analyses les éléments traces appartenant au groupe des terres rares comme
précédemment réalisés au cours de plusieurs études microchimiques (Dorval et al. 2007, Lara
et al. 2008, Tournois et al. 2013). Plusieurs raisons permettent d’avancer leurs contributions à une caractérisation spatiale plus fine du lagon de Nouvelle-Calédonie. Dans un premier
temps, ces éléments sont de bons indicateurs de l’environnement car ils ne sont pas altérés par 168
des processus métaboliques au sein de la masse d’eau. Ainsi, ils sont susceptibles de ne pas être régulés physiologiquement lors de leur incorporation au sein des otolithes et de présenter
des concentrations dans les otolithes reflétant celles de l’environnement (Arslan & Paulson
2003). De plus, ces éléments sont aussi caractéristiques des apports terrigènes provenant des
eaux douces et bassins versants et certains d’entre eux résulteraient des activités anthropiques (indicateur de pollutions ou autres) (Munksgaard et al. 2003, Nozaki 2009).
Certains ratios élémentaires et isotopiques, tels que les ratios Ba/Sr et 87Sr/86Sr, diffèrent
significativement entre le milieu marin et le milieu dulçaquicole (McCulloch et al. 2005,
McMahon et al. 2013). Ils permettraient de retracer les migrations diadromiques en
caractérisant la fréquentation de ces deux milieux ainsi que le passage au sein
d’environnements de transition tels que les estuaires. La Nouvelle-Calédonie présente de
nombreux estuaires le long de sa façade maritime et l’étude de ces ratios pourraient apporter
de nouvelles connaissances sur les patterns de mouvements des poissons diadromiques
calédoniens.
Dans l’optique d’études supplémentaires renseignant sur les possibles usages de la
microchimie des otolithes au sein du paysage marin de la Nouvelle-Calédonie, il serait
intéressant d’établir une première série de pistes quant à la potentialité de l’outil en tant qu’enregistreur d’impacts environnementaux. En effet, la microchimie des otolithes pourrait
révéler les traces de l’influence anthropique passée et actuelle en Nouvelle-Calédonie telles
que l’exploitation minière, le développement des zones urbaines ou des événements chimiques exceptionnels. Ces pistes ont pu être initiéés au cours d’un programme d’étude de la microchimie des otolithes et de l’environnement de Nouvelle-Calédonie (Vigliola et al.
2013). Ce travail n’a pas permis de différencier les sites sous influence anthropique par
l’analyse directe de la microchimie des otolithes de poisson. Cependant, l’outil de
classification utilisé a permis de détecter ces sites en y associant une probabilité d’impact. Les
résultats de cette étude préliminaire ainsi que l’ajout aux analyses microchimiques des terres
rares, qui pour certains résulteraient directement des activités anthropiques, encouragent la
réalisation d’études supplémentaires visant à approfondir le potentiel de l’outil microchimie en tant qu’indicateur d’impact.
169
170
Références bibliographiques
Adams, A. J., C. P. Dahlgren, G. T. Kellison, M. S. Kendall, C. A. Layman, J. A. Ley, I.
Nagelkerken, J. E. Serafy (2006). "Nursery function of tropical back-reef systems." Marine
Ecology-Progress Series 318: 287-301.
Allen, G., R. Steene, P. Humann, N. Deloach (2003). Reef Fish Identification: Tropical
Pacific. Jacksonville, New World Publications.
Almany, G. R., M. L. Berumen, S. R. Thorrold, S. Planes, G. P. Jones (2007). "Local
replenishment of coral reef fish populations in a marine reserve." Science 316(5825): 742744.
Almany, G. R., S. R. Connolly, D. D. Heath, J. D. Hogan, J. P. Jones, L. J. McCook, M.
Mills, R. L. Pressey, D. H. Williamson (2009). "Connectivity, biodiversity conservation and
the design of marine reserve networks for coral reefs." Coral Reefs 28: 339-351.
Alongi, D. M. (2002). "Present state and future of the world's mangrove forests."
Environmental Conservation 29(3): 331-349.
Ambatsian, P., F. Fernex, M. Bernat, C. Parron, J. Lecolle (1997). "High metal inputs to
closed seas: The New Caledonian lagoon." Journal of Geochemical Exploration 59(1): 59-74.
Anderson, M. J. (2001). "A new method for non-parametric multivariate analysis of
variance." Austral Ecology 26(1): 32-46.
Andréfouët, S., G. Cabioch, B. Flamand, B. Pelletier (2009). "A reappraisal of the diversity of
geomorphological and genetic processes of New Caledonian coral reefs: a synthesis from
optical remote sensing, coring and acoustic multibeam observations." Coral Reefs 28(3): 691707.
Andréfouët, S., D. Torres-Pulliza (2004). Atlas des récifs coralliens de Nouvelle-Calédonie.
IFRECOR Nouvelle-Calédonie. IRD, Nouméa: 26 p + 22 planches.
Arai, T., M. Ohji, T. Hirata (2007). "Trace metal deposition in teleost fish otolith as an
environmental indicator." Water Air and Soil Pollution 179(1-4): 255-263.
Arslan, Z., A. J. Paulson (2003). "Solid phase extraction for analysis of biogenic carbonates
by electrothermal vaporization inductively coupled plasma mass spectrometry (ETV-ICPMS): an investigation of rare earth element signatures in otolith microchemistry." Analytica
Chimica Acta 476(1): 1-13.
171
Baille, J. S., F. Colin, D. Marini, G. Monteil, M. Lardy (2012). La mine et le traitement des
minerais de Nickel. Atlas de la Nouvelle-Calédonie. J. Bonvallot. Marseille (FRA), Nouméa,
IRD, Congrès de la Nouvelle-Calédonie: 165-168.
Barnes, T. C., B. M. Gillanders (2013). "Combined effects of extrinsic and intrinsic factors on
otolith chemistry: implications for environmental reconstructions." Canadian Journal of
Fisheries and Aquatic Sciences 70: 1159-1166.
Bath, G. E., S. R. Thorrold, C. M. Jones, S. E. Campana, J. W. McLaren, J. W. H. Lam
(2000). "Strontium and barium uptake in aragonitic otoliths of marine fish." Geochimica Et
Cosmochimica Acta 64(10): 1705-1714.
Beck, M. W., K. L. Heck, K. W. Able, D. L. Childers, D. B. Eggleston, B. M. Gillanders, B.
Halpern, C. G. Hays, K. Hoshino, T. J. Minello, R. J. Orth, P. F. Sheridan, M. P. Weinstein
(2001). "The identification, conservation, and management of estuarine and marine nurseries
for fish and invertebrates." BioScience 51(8): 633-641.
Beck, M. W., K. L. Heck, K. W. Able, D. L. Childers, D. B. Eggleston, B. M. Gillanders, B.
S. Halpern, C. G. Hays, K. Hoshino, T. J. Minello (2003). "The role of nearshore ecosystems
as fish and shellfish nurseries." Issues in Ecology.
Bellwood, D. R., T. P. Hughes, C. Folke, M. Nystrom (2004). "Confronting the coral reef
crisis." Nature 429(6994): 827-833.
Bellwood, R. D., T. P. Hughes (2001). "Regional-scale assembly rules and biodiversity of
corals reefs." Science 292: 1532-1534.
Berkstrom, C., M. Gullstrom, R. Lindborg, A. W. Mwandya, S. A. S. Yahya, N. Kautsky, M.
Nystrom (2012). "Exploring 'knowns' and 'unknowns' in tropical seascape connectivity with
insights from East African coral reefs." Estuarine Coastal and Shelf Science 107: 1-21.
Berkstrom, C., T. L. Jorgensen, M. Hellstrom (2013). "Ecological connectivity and niche
differentiation between two closely related fish species in the mangrove-seagrass-coral reef
continuum." Marine Ecology Progress Series 477: 201-215.
Berumen, M. L., H. J. Walsh, N. Raventos, S. Planes, G. P. Jones, V. Starczak, S. R. Thorrold
(2010). "Otolith geochemistry does not reflect dispersal history of clownfish larvae." Coral
Reefs 29(4): 883-891.
Bonvallot, J., M. Lardy (2012). Un grand complexe métallurgique dans la Province Sud : Vale
Nouvelle-Calédonie. Atlas de la Nouvelle-Calédonie. J. Bonvallot. Marseille (FRA), Nouméa,
IRD, Congrès de la Nouvelle-Calédonie: 173-176.
Breiman, L. (2001). "Random forests." Machine Learning 45(1): 5-32.
172
Brown, J. A. (2006a). "Classification of juvenile flatfishes to estuarine and coastal habitats
based on the elemental composition of otoliths." Estuarine Coastal and Shelf Science 66(3-4):
594-611.
Brown, J. A. (2006b). "Using the chemical composition of otoliths to evaluate the nursery role
of estuaries for English sole Pleuronectes vetulus populations." Marine Ecology Progress
Series 306: 269-281.
Bruland, K. W., M. C. Lohan, H. D. Holland, K. K. Turekian (2003). 6.02 - Controls of Trace
Metals in Seawater. Treatise on Geochemistry. Oxford, Pergamon: 23-47.
Buckel, J. A., B. L. Sharack, V. S. Zdanowicz (2004). "Effect of diet on otolith composition
in Pomatomus saltatrix, an estuarine piscivore." Journal of Fish Biology 64(6): 1469-1484.
Burke, L. M., K. Reytar, M. Spalding, A. Perry (2011). Reefs at risk revisited, World
Resources Institute.
Calabrese, J. M., W. F. Fagan (2004). "A comparison-shopper's guide to connectivity
metrics." Frontiers in Ecology and the Environment 2(10): 529-536.
Campana, S. E. (1999). "Chemistry and composition of fish otoliths: pathways, mechanisms
and applications." Marine Ecology Progress Series 188: 263-297.
Campana, S. E. (2005). Otolith Elemental Composition as a Natural Marker of Fish Stocks.
Stock Identification Methods. Applications in Fishery Science. S. X. Cadrin, K. D. Friedland
and J. R. Waldman, Elsevier Academic Press: 227-245.
Campana, S. E., J. M. Casselman (1993). "Stock discrimination using otolith shape-analysis."
Canadian Journal of Fisheries and Aquatic Sciences 50(5): 1062-1083.
Campana, S. E., G. A. Chouinard, J. M. Hanson, A. Frechet, J. Brattey (2000). "Otolith
elemental fingerprints as biological tracers of fish stocks." Fisheries Research 46(1-3): 343357.
Campana, S. E., J. D. Neilson (1985). "Microstructure of fish otoliths." Canadian Journal of
Fisheries and Aquatic Sciences 42(5): 1014-1032.
Campana, S. E., S. R. Thorrold (2001). "Otoliths, increments, and elements: keys to a
comprehensive understanding of fish populations?" Canadian Journal of Fisheries and
Aquatic Sciences 58(1): 30-38.
Carassou, L., M. Leopold, N. Guillemot, L. Wantiez, M. Kulbicki (2013). "Does herbivorous
fish protection really improve coral reef resilience? A case study from New Caledonia (South
Pacific)." Plos One 8(4): 12.
173
Carpenter, K. E., V. H. Niem (2001a). FAO species identification guide for fishery purposes.
The Living Marine Resources of the Western Central Pacific: Bony fishes part 3 (Menidae to
Pomacentridae). FAO. Rome, FAO Fisheries Department. 5: pp. 2791-3380.
Carpenter, K. E., V. H. Niem (2001b). FAO species identification guide for fishery purposes.
The Living Marine Resources of the Western Central Pacific: Bony fishes part 4 (Labridae to
Latimeriidae), estuarine crocodiles, sea turtles, sea snakes and marine mammals. FAO. Rome,
FAO Fisheries Department. 6: pp. 3381-4218.
Carpenter, S. J., J. M. Erickson, F. D. Holland (2003). "Migration of a Late Cretaceous fish."
Nature 423(6935): 70-74.
Chabanet, P., N. Guillemot, M. Kulbicki, L. Vigliola, S. Sarramegna (2010). "Baseline study
of the spatio-temporal patterns of reef fish assemblages prior to a major mining project in
New Caledonia (South Pacific)." Marine Pollution Bulletin 61(7-12): 598-611.
Chateau, O., L. Wantiez (2009). "Movement patterns of four coral reef fish species in a
fragmented habitat in New Caledonia: implications for the design of marine protected area
networks." Ices Journal of Marine Science 66(1): 50-55.
Chateau, O., L. Wantiez, R. Farman (2012). "Does habitat fragmentation restrict coral reef
fish movements?" 12th International Coral Reef Symposium, ICRS Cairns Poster.
Chittaro, P. M., R. J. Finley, P. S. Levin (2009). "Spatial and temporal patterns in the
contribution of fish from their nursery habitats." Oecologia 160(1): 49-61.
Chittaro, P. M., B. J. Fryer, P. F. Sale (2004). "Discrimination of French grunts (Haemulon
flavolineatum, Desmarest, 1823) from mangrove and coral reef habitats using otolith
microchemistry." Journal of Experimental Marine Biology and Ecology 308: 169-183.
Chittaro, P. M., J. D. Hogan (2013). "Patterns of connectivity among populations of a coral
reef fish." Coral Reefs 32(2): 341-354.
Chittaro, P. M., P. Usseglio, B. J. Fryer, P. F. Sale (2005). "Using otolith microchemistry of
Haemulon flavolineatum (French Grunt) to characterize mangroves and coral reefs throughout
Turneffe Atoll, Belize: Difficulties at small spatial scales." Estuaries 28(3): 373-381.
Chittaro, P. M., P. Usseglio, B. J. Fryer, P. F. Sale (2006). "Spatial variation in otolith
chemistry of Lutjanus apodus at Turneffe Atoll, Belize." Estuarine Coastal and Shelf Science
67(4): 673-680.
Choat, J. H., D. R. Robertson (2001). Age-Based Studies on Coral Reef Fishes. Coral Reef
Fishes, Academic Press: 57-80.
Choat, J. H., D. R. Robertson, J. L. Ackerman, J. M. Posada (2003). "An age-based
demographic analysis of the Caribbean stoplight parrotfish Sparisoma viride." Marine
174
Clausen, L. A. W., D. Bekkevold, E. M. C. Hatfield, H. Mosegaard (2007). "Application and
validation of otolith microstructure as a stock identification method in mixed Atlantic herring
(Clupea harengus) stocks in the North Sea and western Baltic." Ices Journal of Marine
Science 64(2): 377-385.
Claydon, J. A. B., M. I. McCormick, G. P. Jones (2012). "Patterns of migration between
feeding and spawning sites in a coral reef surgeonfish." Coral Reefs 31(1): 77-87.
Cocheret de la Morinière, E., B. J. A. Pollux, I. Nagelkerken, M. A. Hemminga, A. H. L.
Huiskes, G. van der Velde (2003a). "Ontogenetic dietary changes of coral reef fishes in the
mangrove-seagrass-reef continuum: stable isotopes and gut-content analysis." Marine
Cocheret de la Morinière, E., B. J. A. Pollux, I. Nagelkerken, G. van der Velde (2003b). "Diet
shifts of Carribean grunts (Haemulidae) and snappers (Lutjanidae) and the relation with
nursery-to-coral reef migrations." Estuarine, Coastal and Shelf Science 57: 1079-1089.
Cowen, R. K., G. Gawarkiewicz, J. Pineda, S. R. Thorrold, F. E. Werner (2007). "Population
connectivity in marine systems, an overview." Oceanography 20(3): 14-21.
Cowen, R. K., K. M. M. Lwiza, S. Sponaugle, C. B. Paris, D. B. Olson (2000). "Connectivity
of marine populations: open or closed?" Science 287: 857-859.
Cowen, R. K., C. B. Paris, A. Srinivasan (2006). "Scaling of connectivity in marine
populations." Science 311: 522-527.
Cowen, R. K., S. Sponaugle (2009). "Larval dispersal and marine population connectivity."
Annual Review of Marine Science 1: 443-466.
Crooks, K. R., M. Sanjayan (2006). Connectivity conservation: maintening connections for
nature. Connectivity Conservation. K. R. Crooks and M. Sanjayan. New York, Cambridge
University Press: 1-20.
Cuveliers, E. L., A. J. Geffen, J. Guelinckx, J. A. M. Raeymaekers, J. Skadal, F. A. M.
Volckaert, G. E. Maes (2010). "Microchemical variation in juvenile Solea solea otoliths as a
powerful tool for studying connectivity in the North Sea." Marine Ecology-Progress Series
401: 211-220.
D’Agata, S., D. Mouillot, M. Kulbicki, S. Andrefouet, D. Bellwood, J. E. Cinner, P. F.
Cowman, M. Kronen, S. Pinca, L. Vigliola (In press). "Human-mediated loss of phylogenetic
and functional diversity in coral reef fishes." Current Biology.
Dahlgren, C. P., G. T. Kellison, A. J. Adams, B. M. Gillanders, M. S. Kendall, C. A. Layman,
J. A. Ley, I. Nagelkerken, J. E. Serafy (2006). "Marine nurseries and effective juvenile
habitats: concepts and applications." Marine Ecology-Progress Series 312: 291-295.
175
De Pontual, H., F. Lagardere, R. Amara, M. Bohn, A. Ogor (2003). "Influence of ontogenetic
and environmental changes in the otolith microchemistry of juvenile sole (Solea solea)."
Journal of Sea Research 50(2-3): 199-210.
Di Franco, A., B. M. Gillanders, G. De Benedetto, A. Pennetta, G. A. De Leo, P. Guidetti
(2012). "Dispersal patterns of coastal fish: implications for designing networks of Marine
Protected Areas." Plos One 7(2): 9.
Disspain, M., L. A. Wallis, B. M. Gillanders (2011). "Developing baseline data to understand
environmental change: a geochemical study of archaeological otoliths from the Coorong,
South Australia." Journal of Archaeological Science 38(8): 1842-1857.
Domeier, M. L., P. L. Colin (1997). "Tropical reef fish spawning aggregations: Defined and
reviewed." Bulletin of Marine Science 60(3): 698-726.
Dorenbosch, M., M. G. G. Grol, M. J. A. Christianen, I. Nagelkerken, G. van der Velde
(2005a). "Indo-Pacific seagrass beds and mangroves contribute to fish density coral and
diversity on adjacent reefs." Marine Ecology Progress Series 302: 63-76.
Dorenbosch, M., M. G. G. Grol, I. Nagelkerken, G. van der Velde (2005b). "Distribution of
coral reef fishes along a coral reef-seagrass gradient: edge effects and habitat segregation."
Marine Ecology-Progress Series 299: 277-288.
Dorenbosch, M., M. G. G. Grol, I. Nagelkerken, G. van der Velde (2006). "Seagrass beds and
mangroves as potential nurseries for the threatened Indo-Pacific humphead wrasse, Cheilinus
undulatus and Caribbean rainbow parrotfish, Scarus quacamaia." Biological Conservation
129(2): 277-282.
Dorenbosch, M., M. Verweij, I. Nagelkerken, N. Jiddawi, G. van der Velde (2004). "Homing
and daytime tidal movements of juvenile snappers (Lutjanidae) between shallow-water
nursery habitats in Zanzibar, Western Indian Ocean." Environmental Biology of Fishes 70(3):
203-209.
Dorval, E., C. M. Jones, R. Hannigan, J. van Montfrans (2007). "Relating otolith chemistry to
surface water chemistry in a coastal plain estuary." Canadian Journal of Fisheries and Aquatic
Sciences 64: 411-424.
Dove, S. G., M. J. Kingsford (1998). "Use of otoliths and eye lenses for measuring tracemetal incorporation in fishes: a biogeographic study." Marine Biology 130(3): 377-387.
Duke, N. C., J. O. Meynecke, S. Dittmann, A. M. Ellison, K. Anger, U. Berger, S. Cannicci,
K. Diele, K. C. Ewel, C. D. Field, N. Koedam, S. Y. Lee, C. Marchand, I. Nordhaus, F.
Dahdouh-Guebas (2007). "A world without mangroves?" Science 317(5834): 41-42.
Elsdon, T. S., B. M. Gillanders (2002). "Interactive effects of temperature and salinity on
otolith chemistry: challenges for determining environmental histories of fish." Canadian
Journal of Fisheries and Aquatic Sciences 59(11): 1796-1808.
176
Elsdon, T. S., B. M. Gillanders (2003a). "Reconstructing migratory patterns of fish based on
environmental influences on otolith chemistry." Reviews in Fish Biology and Fisheries 13(3):
219-235.
Elsdon, T. S., B. M. Gillanders (2003b). "Relationship between water and otolith elemental
concentrations in juvenile black bream Acanthopagrus butcheri." Marine Ecology-Progress
Series 260: 263-272.
Elsdon, T. S., B. M. Gillanders (2004). "Fish otolith chemistry influenced by exposure to
multiple environmental variables." Journal of Experimental Marine Biology and Ecology
313(2): 269-284.
Elsdon, T. S., B. K. Wells, S. E. Campana, B. M. Gillanders, C. M. Jones, K. E. Limburg, D.
H. Secor, S. R. Thorrold, B. D. Walther (2008). "Otolith chemistry to describe movements
and life-history parameters of fishes: Hypotheses, assumptions, limitations and inferences."
Oceanography and Marine Biology: An Annual Review, Vol 46 46: 297-+.
Fairclough, D. V., J. S. Edmonds, R. C. J. Lenanton, G. Jackson, I. S. Keay, B. M. Crisafulli,
S. J. Newman (2011). "Rapid and cost-effective assessment of connectivity among
assemblages of Choerodon rubescens (Labridae), using laser ablation ICP-MS of sagittal
otoliths." Journal of Experimental Marine Biology and Ecology 403(1-2): 46-53.
Ferguson, G. J., T. M. Ward, B. M. Gillanders (2011). "Otolith shape and elemental
composition: Complementary tools for stock discrimination of mulloway (Argyrosomus
japonicus) in southern Australia." Fisheries Research 110(1): 75-83.
Fernandez, J. M., S. Ouillon, C. Chevillon, P. Douillet, R. Fichez, R. Le Gendre (2006). "A
combined modelling and geochemical study of the fate of terrigenous inputs from mixed
natural and mining sources in a coral reef lagoon (New Caledonia)." Marine Pollution
Bulletin 52(3): 320-331.
Fichez, R., L. Breau, C. Chevillon, S. Chifflet, P. Douillet, V. Faure, J. M. Fernandez, P.
Gérard, L. Hédouin, A. Lapetite, S. Ouillon, O. Pringault, J. P. Torréton (2008). "Origine,
transport et devenir des apports naturels et anthropiques dans le lagon sud-ouest de NouvelleCalédonie." Le Journal de la Société des Océanistes 126-127: 1-23.
Fisher, R., D. R. Bellwood, S. D. Job (2000). "Development of swimming abilities in reef fish
larvae." Marine Ecology-Progress Series 202: 163-173.
Forrester, G. E., S. E. Swearer (2002). "Trace elements in otoliths indicate the use of opencoast versus bay nursery habitats by juvenile California halibut." Marine Ecology Progress
Series 241: 201-213.
Fowler, A. J., S. E. Campana, C. M. Jones, S. R. Thorrold (1995). "Experimental assessment
of the effect of temperature and salinity on elemental composition of otoliths using solutionbased ICPMS." Canadian Journal of Fisheries and Aquatic Sciences 52(7): 1421-1430.
177
Geffen, A. J., N. J. G. Pearce, W. T. Perkins (1998). "Metal concentrations in fish otoliths in
relation to body composition after laboratory exposure to mercury and lead." Marine Ecology
Progress Series 165: 235-245.
Gillanders, B. M. (2002). "Connectivity between juvenile and adult fish populations: do adults
remain near their recruitment estuaries?" Marine Ecology-Progress Series 240: 215-223.
Gillanders, B. M. (2005). "Using elemental chemistry of fish otoliths to determine
connectivity between estuarine and coastal habitats." Estuarine, Coastal and Shelf Science 64:
47-57.
Gillanders, B. M. (2009). Tools for Studying Biological Marine Ecosystem Interactions Natural and Artificial Tags. Ecological Connectivity among Tropical Coastal Ecosystems. I.
Nagelkerken, Springer: 457-492.
Gillanders, B. M., K. W. Able, J. A. Brown, D. B. Eggleston, P. F. Sheridan (2003).
"Evidence of connectivity between juvenile and adult habitats for mobile marine fauna: an
important component of nurseries." Marine Ecology Progress Series 247: 281-295.
Gillanders, B. M., M. J. Kingsford (2000). "Elemental fingerprints of otoliths of fish may
distinguish estuarine "nursery" habitats." Marine Ecology Progress Series 201: 273-286.
Gillanders, B. M., M. J. Kingsford (2003). "Spatial variation in elemental composition of
otoliths of three species of fish (family Sparidae)." Estuarine Coastal and Shelf Science 57(56): 1049-1064.
Gillanders, B. M., P. Sanchez-Jerez, J. Bayle-Sempere, A. Ramos-Espla (2001). "Trace
elements in otoliths of the two-banded bream from a coastal region in the south-west
Mediterranean: are there differences among locations?" Journal of Fish Biology 59(2): 350363.
Grandcourt, E. M., T. Z. Al Abdessalaam, F. Francis (2006). "Age, growth, mortality and
reproduction of the blackspot snapper, Lutjanus fulviflamma (Forsskal, 1775), in the southern
Arabian Gulf." Fisheries Research 78(2-3): 203-210.
Grenon, Y., M. Simard (2012). Un grand projet métallurgique dans la Province Nord :
Koniambo. Atlas de la Nouvelle-Calédonie. J. Bonvallot. Marseille (FRA), Nouméa, IRD,
Congrès de la Nouvelle-Calédonie: 177-180.
Grüss, A., D. M. Kaplan, S. Guenette, C. M. Roberts, L. W. Botsford (2011). "Consequences
of adult and juvenile movement for marine protected areas." Biological Conservation 144(2):
692-702.
Guillemot, N., M. Kulbicki, P. Chabanet, L. Vigliola (2011). "Functional redundancy patterns
reveal non-random assembly rules in a species-rich marine assemblage." Plos One 6(10):
e26735.
178
Hamer, P. A., G. P. Jenkins (2007). "Comparison of spatial variation in otolith chemistry of
two fish species and relationships with water chemistry and otolith growth." Journal of Fish
Biology 71(4): 1035-1055.
Hamilton, S. L., J. Regetz, R. R. Warner (2008). "Postsettlement survival linked to larval life
in a marine fish." Proceedings of the National Academy of Sciences of the United States of
America 105(5): 1561-1566.
Hammerschlag-Peyer, C. M., C. A. Layman (2010). "Intrapopulation variation in habitat use
by two abundant coastal fish species." Marine Ecology Progress Series 415: 211-220.
Hansen, L. P., J. A. Jacobsen (2003). "Origin and migration of wild and escaped farmed
Atlantic salmon, Salmo salar L., in oceanic areas north of the Faroe Islands." Ices Journal of
Marine Science 60(1): 110-119.
Hastein, T., B. J. Hill, F. Berthe, D. V. Lightner (2001). "Traceability of aquatic animals."
Revue Scientifique et Technique de l'Office International des Epizooties 20(2): 564-583.
Hedgecock, D., P. H. Barber, S. Edmands (2007). "Genetic Approaches to Measuring
Connectivity." Oceanography 20(3): 70-79.
Heinrich, C. A., T. Pettke, W. E. Halter, M. Aigner-Torres, A. Audetat, D. Gunther, B.
Hattendorf, D. Bleiner, M. Guillong, I. Horn (2003). "Quantitative multi-element analysis of
minerals, fluid and melt inclusions by laser-ablation inductively-coupled-plasma massspectrometry." Geochimica Et Cosmochimica Acta 67(18): 3473-3497.
Hicks, A. S., G. P. Closs, S. E. Swearer (2010). "Otolith microchemistry of two
amphidromous galaxiids across an experimental salinity gradient: A multi-element approach
for tracking diadromous migrations." Journal of Experimental Marine Biology and Ecology
394(1-2): 86-97.
Hitt, S., S. J. Pittman, R. S. Nemeth (2011). "Diel movements of fishes linked to benthic
seascape structure in a Caribbean coral reef ecosystem." Marine Ecology Progress Series 427:
275-291.
Hughes, T. P., A. H. Baird, D. R. Bellwood, M. Card, S. R. Connolly, C. Folke, R. Grosberg,
O. Hoegh-Guldberg, J. B. C. Jackson, J. Kleypas, J. M. Lough, P. Marshall, M. Nystrom, S.
R. Palumbi, J. M. Pandolfi, B. Rosen, J. Roughgarden (2003). "Climate change, human
impacts, and the resilience of coral reefs." Science 301(5635): 929-933.
Hughes, T. P., N. A. J. Graham, J. B. C. Jackson, P. J. Mumby, R. S. Steneck (2010). "Rising
to the challenge of sustaining coral reef resilience." Trends in Ecology & Evolution 25(11):
633-642.
Igulu, M. M., I. Nagelkerken, R. Fraaije, R. van Hitum, H. Ligtenberg, Y. D. Mgaya (2011).
"The potential role of visual cues for microhabitat selection during the early life phase of a
179
coral reef fish (Lutjanus fulviflamma)." Journal of Experimental Marine Biology and Ecology
401(1-2): 118-125.
Jones, D. L., J. F. Walter, E. N. Brooks, J. E. Serafy (2010). "Connectivity throught ontogeny:
fish population linkages among mangroves and coral reef habitats." Marine Ecology Progress
Series 401: 245-258.
Jones, G. P., G. R. Almany, G. R. Russ, P. F. Sale, R. S. Steneck, M. J. H. van Oppen, B. L.
Willis (2009). "Larval retention and connectivity among populations of corals and reef fishes:
history, advances and challenges." Coral Reefs 28: 307-325.
Jones, G. P., M. J. Milicich, M. J. Emslie, C. Lunow (1999). "Self-recruitment in a coral reef
fish population." Nature 402(6763): 802-804.
Jones, G. P., S. Planes, S. R. Thorrold (2005). "Coral reef fish larvae settle close to home."
Current Biology 15(14): 1314-1318.
Jones, G. P., M. Srinivasan, G. R. Almany (2007). "Population Connectivity and
Conservation of Marine Biodiversity." Oceanography 20(3): 100-111.
Kamukuru, A. T., Y. D. Mgaya (2004). "The food and feeding habits of blackspot snapper,
Lutjanus fulviflamma (Pisces : Lutjanidae) in shallow waters of Mafia Island, Tanzania."
African Journal of Ecology 42(1): 49-58.
Kaunda-Arara, B., G. A. Rose (2004a). "Long-distance movements of coral reef fishes." Coral
Reefs 23(3): 410-412.
Kaunda-Arara, B., G. A. Rose (2004b). "Out-migration of tagged fishes from marine reef
National Parks to fisheries in coastal Kenya." Environmental Biology of Fishes 70(4): 363372.
Kennedy, B. P., J. D. Blum, C. L. Folt, K. H. Nislow (2000). "Using natural strontium
isotopic signatures as fish markers: methodology and application." Canadian Journal of
Kimirei, I. A., I. Nagelkerken, B. Griffioen, C. Wagner, Y. D. Mgaya (2011). "Ontogenetic
habitat use by mangrove/seagrass-associated coral reef fishes shows flexibility in time and
space." Estuarine Coastal and Shelf Science 92(1): 47-58.
Kingsford, M. J., B. M. Gillanders (2000). "Variation in concentrations of trace elements in
otoliths and eye lenses of a temperate reef fish, Parma microlepis, as a function of depth,
spatial scale, and age." Marine Biology 137(3): 403-414.
Kingsford, M. J., J. M. Leis, A. Shanks, K. C. Lindeman, S. G. Morgan, J. Pineda (2002).
"Sensory environments, larval abilities and local self-recruitment." Bulletin of Marine Science
70(1): 309-340.
180
Kramer, D. L., M. R. Chapman (1999). "Implications of fish home range size and relocation
for marine reserve function." Environmental Biology of Fishes 55(1-2): 65-79.
Kulbicki, M. (2006). Ecologie des poissons lagonaires de Nouvelle Calédonie. Milieux,
Organismes et Evolution. Perpignan, EPHE à l'Université de Perpignan: 195.
Kulbicki, M., Y. M. Bozec, P. Labrosse, Y. Letourneur, G. Mou-Tham, L. Wantiez (2005a).
"Diet composition of carnivorous fishes from coral reef lagoons of New Caledonia." Aquatic
Living Resources 18(3): 231-250.
Kulbicki, M., N. Guillemot, M. Amand (2005b). "A general approach to length-weight
relationships for New Caledonian lagoon fishes." Cybium 29(3): 235-252.
Kulbicki, M., J. Rivaton (1997). "Inventaire et biogéographie des poissons lagonaires et
récifaux de Nouvelle-Calédonie." Cybium 21 (1Suppl.): 81-98.
L'Huillier, L., T. Jaffré, A. Wulff (2010). Mines et Environnement en Nouvelle-Calédonie: les
milieux sur substrats ultramafiques et leur restauration. Nouméa, Nouvelle-Calédonie.
Labonne, M., E. Morize, M. Kulbicki, D. Ponton, L. Marec (2008). "Otolith chemical
signature and growth of Chaetodon speculum in coastal areas of New Caledonia." Estuarine,
Coastal and Shelf Science 78: 493-504.
Laboute, P., R. Grandperrin (2009). Guide des poissons de Nouvelle-Calédonie. Nouméa,
Editions Catherine Ledru.
Labrosse, P., R. Fichez, R. Farman, T. Adams (2000). New Caledonia. Seas at the Millenium,
an Environmental Evaluation. Elsevier. Amsterdam. 2: 723-736.
Labrosse, P., M. Kulbicki, J. Ferraris (2002). Underwater Visual Fish Census Surveys. Proper
Use and Implementation. Nouméa, New Caledonia.
Laegdsgaard, P., C. Johnson (2001). "Why do juvenile fish utilise mangrove habitats?"
Journal of Experimental Marine Biology and Ecology 257: 229-253.
Lara, M. R., D. L. Jones, Z. Chen, J. T. Lamkin, C. M. Jones (2008). "Spatial variation of
otolith elemental signatures among juvenile gray snapper (Lutjanus griseus) inhabiting
southern Florida waters." Marine Biology 153(3): 235-248.
Layman, C. A., C. P. Dahlgren, G. T. Kellison, A. J. Adams, B. M. Gillanders, M. S. Kendall,
J. A. Ley, I. Nagelkerken, J. E. Serafy (2006). "Marine nurseries and effective juvenile
habitats." Marine Ecology-Progress Series 318: 307-308.
Lea, E. (1910). "On the methods used in the herring investigations." Publications de
Circonstance. Conseil Permanent International pour l'Exploration de la Mer. Copenhague. 53:
7-25.
181
Lecomte-Finiger, R. (1992). "Growth history and age at recruitment of European glass eels
(Anguilla anguilla) as revealed by otolith microstructure." Marine Biology 114(2): 205-210.
Legendre, P., M. J. Anderson (1999). "Distance-based redundancy analysis: Testing
multispecies responses in multifactorial ecological experiments." Ecological Monographs
69(1): 1-24.
Leis, J. M., K. J. Wright, R. N. Johnson (2007). "Behaviour that influences dispersal and
connectivity in the small, young larvae of a reef fish." Marine Biology 153(1): 103-117.
Lester, S. E., B. I. Ruttenberg (2005). "The relationship between pelagic larval duration and
range size in tropical reef fishes: a synthetic analysis." Proceedings of the Royal Society BBiological Sciences 272(1563): 585-591.
Levin, L. A. (2006). "Recent progress in understanding larval dispersal: new directions and
digressions." Integrative and Comparative Biology 46(3): 282.
Lin, S. H., C. W. Chang, Y. Iizuka, W. N. Tzeng (2007). "Salinities, not diets, affect
strontium/calcium ratios in otoliths of Anguilla japonica." Journal of Experimental Marine
Biology and Ecology 341(2): 254-263.
Lo-Yat, A., M. Meekan, N. Munksgaard, D. Parry, S. Planes, M. Wolter, J. Carleton (2005).
"Small-scale spatial variation in the elemental composition of otoliths of Stegastes nigricans
(Pomacentridae) in French Polynesia." Coral Reefs 24(4): 646-653.
Longerich, H. P., S. E. Jackson, D. Gunther (1996). "Laser ablation inductively coupled
plasma mass spectrometric transient signal data acquisition and analyte concentration
calculation." Journal of Analytical Atomic Spectrometry 11(9): 899-904.
Loubens, G. (1980). "Biologie de quelques espèces de Poissons du lagon neo-Calédonien. II.
Sexualité et reproduction." Cahiers de l’Indo-pacifique Volume 2(1): 41-72.
Lowe, W. H., F. W. Allendorf (2010). "What can genetics tell us about population
connectivity?" Molecular Ecology 19(15): 3038-3051.
Lugendo, B. R., I. Nagelkerken, G. Kruitwagen, G. van der Velde, Y. D. Mgaya (2007).
"Relative importance of mangroves as feeding habitats for fishes: a comparison between
mangrove habitats with different settings." Bulletin of Marine Science 80(3): 497-512.
Lugendo, B. R., I. Nagelkerken, G. van der Velde, Y. D. Mgaya (2006). "The importance of
mangroves, mud and sand flats, and seagrass beds as feeding areas for juvenile fishes in
Chwaka Bay, Zanzibar: gut content and stable isotope analyses." Journal of Fish Biology
69(6): 1639-1661.
Lugendo, B. R., A. Pronker, I. Cornelissen, A. de Groene, I. Nagelkerken, M. Dorenbosch, G.
van der Velde, Y. D. Mgaya (2005). "Habitat utilisation by juveniles of commercially
182
important fish species in a marine embayment in Zanzibar, Tanzania." Aquatic Living
Resources 18(2): 149-158.
Luiz, O. J., A. P. Allen, D. R. Robertson, S. R. Floeter, M. Kulbicki, L. Vigliola, R. Becheler,
J. S. Madin (2013). "Adult and larval traits as determinants of geographic range size among
tropical reef fishes." Proceedings of the National Academy of Sciences of the United States of
America 110(41): 16498-16502.
Luiz, O. J., J. S. Madin, D. R. Robertson, L. A. Rocha, P. Wirtz, S. R. Floeter (2012).
"Ecological traits influencing range expansion across large oceanic dispersal barriers: insights
from tropical Atlantic reef fishes." Proceedings of the Royal Society B-Biological Sciences
279(1730): 1033-1040.
Macpherson, E., N. Raventos (2006). "Relationship between pelagic larval duration and
geographic distribution of Mediterranean littoral fishes." Marine Ecology Progress Series 327:
257-265.
Manel, S., O. E. Gaggiotti, R. S. Waples (2005). "Assignment methods: matching biological
questions with appropriate techniques." Trends in Ecology & Evolution 20(3): 136-142.
Marchand, C., M. Allenbach, E. Lallier-Verges (2011). "Relationships between heavy metals
distribution and organic matter cycling in mangrove sediments (Conception Bay, New
Caledonia)." Geoderma 160(3-4): 444-456.
Marohn, L., E. Prigge, K. Zumholz, A. Klugel, H. Anders, R. Hanel (2009). "Dietary effects
on multi-element composition of European eel (Anguilla anguilla) otoliths." Marine Biology
156(5): 927-933.
Martin, G. B., S. R. Thorrold (2005). "Temperature and salinity effects on magnesium,
manganese, and barium incorporation in otoliths of larval and early juvenile spot Leiostomus
xanthurus." Marine Ecology-Progress Series 293: 223-232.
Martin, G. B., M. J. Wuenschel (2006). "Effect of temperature and salinity on otolith element
incorporation in juvenile gray snapper Lutjanus griseus." Marine Ecology-Progress Series
324: 229-239.
Martin, J., G. Bareille, S. Berail, C. Pecheyran, F. Daverat, N. Bru, H. Tabouret, O. Donard
(2013). "Spatial and temporal variations in otolith chemistry and relationships with water
chemistry: a useful tool to distinguish Atlantic salmon Salmo salar parr from different natal
streams." Journal of Fish Biology 82(5): 1556-1581.
Mateo, I., E. G. Durbin, R. S. Appeldoorn, A. J. Adams, F. Juanes, R. Kingsley, P. Swart, D.
Durant (2010). "Role of mangroves as nurseries for French grunt Haemulon flavolineatum
and schoolmaster Lutjanus apodus assessed by otolith elemental fingerprints." Marine
183
McCook, L. J., G. R. Almany, M. L. Berumen, J. C. Day, A. L. Green, G. P. Jones, J. M. Leis,
S. Planes, G. R. Russ, P. F. Sale, S. R. Thorrold (2009). "Management under uncertainty:
guide-lines for incorporating connectivity into the protection of coral reefs." Coral Reefs
28(2): 353-366.
McCulloch, M., M. Cappo, J. Aumend, W. Muller (2005). "Tracing the life history of
individual barramundi using laser ablation MC-ICP-MS Sr-isotopic and Sr/Ba ratios in
otoliths." Marine and Freshwater Research 56(5): 637-644.
McMahon, K. W., M. L. Berumen, I. Mateo, T. S. Elsdon, S. R. Thorrold (2011). "Carbon
isotopes in otolith amino acids identify residency of juvenile snapper (Family: Lutjanidae) in
coastal nurseries." Coral Reefs 30(4): 1135-1145.
McMahon, K. W., M. L. Berumen, S. R. Thorrold (2012). "Linking habitat mosaics and
connectivity in a coral reef seascape." Proceedings of the National Academy of Sciences of
the United States of America 109(38): 15372-15376.
McMahon, K. W., L. L. Hamady, S. R. Thorrold (2013). "A review of ecogeochemistry
approaches to estimating movements of marine animals." Limnology and Oceanography
58(2): 697-714.
Mellin, C., M. Kulbicki, D. Ponton (2007). "Seasonal and ontogenetic patterns of habitat use
in coral reef fish juveniles." Estuarine, Coastal and Shelf Science 75(4): 481-491.
Mercier, L., A. M. Darnaude, O. Bruguier, R. P. Vasconcelos, H. N. Cabral, M. J. Costa, M.
Lara, D. L. Jones, D. Mouillot (2011). "Selecting statistical models and variable combinations
for optimal classification using otolith microchemistry." Ecological Applications 21(4): 13521364.
Mercier, L., D. Mouillot, O. Bruguier, L. Vigliola, A. M. Darnaude (2012). "Multi-element
otolith fingerprints unravel sea-lagoon lifetime migrations of gilthead sea bream Sparus
aurata." Marine Ecology Progress Series 444: 175-194.
Meyer, C. G., Y. P. Papastamatiou, T. B. Clark (2010). "Differential movement patterns and
site fidelity among trophic groups of reef fishes in a Hawaiian marine protected area." Marine
Biology 157(7): 1499-1511.
Migon, C., S. Ouillon, X. Mari, E. Nicolas (2007). "Geochemical and hydrodynamic
constraints on the distribution of trace metal concentrations in the lagoon of Noumea, New
Caledonia." Estuarine Coastal and Shelf Science 74(4): 756-765.
Miller, J. A. (2009). "The effects of temperature and water concentration on the otolith
incorporation of barium and manganese in black rockfish Sebastes melanops." Journal of Fish
Biology 75(1): 39-60.
184
Miller, J. A. (2011). "Effects of water temperature and barium concentration on otolith
composition along a salinity gradient: Implications for migratory reconstructions." Journal of
Experimental Marine Biology and Ecology 405(1-2): 42-52.
Milton, D., I. Halliday, M. Sellin, R. Marsh, J. Staunton-Smith, J. Woodhead (2008). "The
effect of habitat and environmental history on otolith chemistry of barramundi Lates
calcarifer in estuarine populations of a regulated tropical river." Estuarine Coastal and Shelf
Science 78(2): 301-315.
Milton, D. A., S. R. Chenery (2001). "Sources and uptake of trace metals in otoliths of
juvenile barramundi (Lates calcarifer)." Journal of Experimental Marine Biology and
Ecology 264(1): 47-65.
Moilanen, A., I. Hanski (2001). "On the use of connectivity measures in spatial ecology."
Oikos 95(1): 147-151.
Moilanen, A., I. Hanski (2006). Connectivity and metapopulation dynamics in highly
fragmented landscapes. Connectivity Conservation. K. R. Crooks and M. Sanjayan.
Cambridge, Cambridge University Press: 44-71.
Mora, C., P. M. Chittaro, P. F. Sale, J. P. Kritzer, S. A. Ludsin (2003). "Patterns and processes
in reef fish diversity." Nature 421(6926): 933-936.
Mora, C., P. F. Sale (2011). "Ongoing global biodiversity loss and the need to move beyond
protected areas: a review of the technical and practical shortcomings of protected areas on
land and sea." Marine Ecology-Progress Series 434: 251-266.
Morales-Nin, B., A. J. Geffen, F. Cardona, C. Kruber, F. Saborido-Rey (2007). "The effect of
Prestige oil ingestion on the growth and chemical composition of turbot otoliths." Marine
Pollution Bulletin 54(11): 1732-1741.
Mumby, P. J., A. J. Edwards, J. E. Arias-Gonzalez, K. C. Lindeman, P. G. Blackwell, A. Gall,
M. I. Gorczynska, A. R. Harborne, C. L. Pescod, H. Renken, C. C. C. Wabnitz, G. Llewellyn
(2004). "Mangroves enhance the biomass of coral reef fish communities in the Caribbean."
Nature 427: 533-536.
Mumby, P. J., A. Hastings (2008). "The impact of ecosystem connectivity on coral reef
resilience." Journal of Applied Ecology 45: 854-862.
Munday, P. L., S. K. Wilson (1997). "Comparative efficacy of clove oil and other chemicals
in anaesthetization of Pomacentrus amboinensis, a coral reef fish." Journal of Fish Biology
51(5): 931-938.
Munksgaard, N. C., K. Lim, D. L. Parry (2003). "Rare earth elements as provenance
indicators in North Australian estuarine and coastal marine sediments." Estuarine, Coastal and
Shelf Science 57: 399-409.
185
Nagelkerken, I. (2009). Ecological Connectivity among Tropical Coastal Ecosystems,
Springer.
Nagelkerken, I., S. J. M. Blaber, S. Bouillon, P. Green, M. Haywood, L. G. Kirton, J. O.
Meynecke, J. Pawlik, H. M. Penrose, A. Sasekumar, P. J. Somerfield (2008). "The habitat
function of mangroves for terrestrial and marine fauna: A review." Aquatic Botany 89(2):
155-185.
Nagelkerken, I., M. Dorenbosch, W. Verberk, E. Cocheret de la Moriniere, G. van Der Velde
(2000). "Importance of shallow-water biotopes of a Caribbean bay for juvenile coral reef
fishes: patterns in biotope association, community structure and spatial distribution." Marine
Ecology Progress Series 202: 175-192.
Nagelkerken, I., C. M. Roberts, G. van der Velde, M. Dorenbosch, M. C. van Riel, E. C. de la
Morinere, P. H. Nienhuis (2002). "How important are mangroves and seagrass beds for coralreef fish? The nursery hypothesis tested on an island scale." Marine Ecology-Progress Series
244: 299-305.
Nakamura, Y., M. Horinouchi, T. Shibuno, Y. Tanaka, T. Miyajima, I. Koike, H. Kurokura,
M. Sano (2008). "Evidence of ontogenetic migration from mangroves to coral reefs by blacktail snapper Lutjanus fulvus: stable isotope approach." Marine Ecology Progress Series 355:
257-266.
Nations, F. a. A. O. o. t. U. (2007). The world's mangroves 1980-2005. Rome, FAO.
Newman, S. J., I. J. Dunk (2002). "Growth, age validation, mortality, and other population
characteristics of the red emperor snapper, Lutjanus sebae (Cuvier, 1828), off the Kimberley
coast of north-western Australia." Estuarine Coastal and Shelf Science 55(1): 67-80.
Newman, S. J., D. M. Williams, G. R. Russ (1996). "Age validation, growth and mortality
rates of the tropical snappers (Pisces: Lutjanidae) Lutjanus adetii (Castelnau, 1873) and L.
quinquelineatus (Bloch, 1790) from the central great barrier reef, Australia." Marine and
Freshwater Research 47(4): 575-584.
Nozaki, Y. (2009). Rare Earth Elements and Their Isotopes. Encyclopedia of Ocean Sciences
(Second Edition). J. H. Steele, S. A. Thorpe and K. K. Turekian. Oxford, Academic Press. 4:
653-665.
Nystrom, M., C. Folke, F. Moberg (2000). "Coral reef disturbance and resilience in a humandominated environment." Trends in Ecology & Evolution 15(10): 413-417.
Olds, A. D., R. M. Connolly, K. A. Pitt, P. S. Maxwell (2012). "Primacy of seascape
connectivity effects in structuring coral reef fish assemblages." Marine Ecology Progress
Series 462: 191-203.
Paillon, C., L. Wantiez, M. Labonne, L. Vigliola (In prep.). "Assessing the power of otoliths
multi-elemental signatures in the coral reefs and mangroves of New Caledonia."
186
Palumbi, S. R. (2003). "Population genetics, demographic connectivity, and the design of
marine reserves." Ecological Applications 13(sp1): 146-158.
Pandolfi, J. M., R. H. Bradbury, E. Sala, T. P. Hughes, K. A. Bjorndal, R. G. Cooke, D.
McArdle, L. McClenachan, M. J. H. Newman, G. Paredes, R. R. Warner, J. B. C. Jackson
(2003). "Global trajectories of the long-term decline of coral reef ecosystems." Science
301(5635): 955-958.
Panfili, J., H. d. Pontual, H. Troadec, P. J. Wright (2002). Manuel de sclérochronologie des
poissons, Coédition Ifremer-IRD.
Paris, J. P. (1981). Géologie de la Nouvelle-Calédonie. Orléans, Editions du B.R.G.M.
Parrish, J. D. (1989). "Fish communities of interacting shallow-water habitats in tropical
oceanic regions." Marine Ecology-Progress Series 58(1-2): 143-160.
Patterson, H. M., M. J. Kingsford (2005). "Elemental signatures of Acanthochromis
polyacanthus otoliths from the Great Barrier Reef have significant temporal, spatial, and
between-brood variation." Coral Reefs 24(3): 360-369.
Patterson, H. M., M. J. Kingsford, M. T. McCulloch (2004). "Elemental signatures of
Pomacentrus coelestis otoliths at multiple spatial scales on the Great Barrier Reef, Australia."
Marine Ecology-Progress Series 270: 229-239.
Pelletier, B., F. Juffroy, B. Flamand, J. Perrier (2012). La bathymétrie des marges de la
Grande Terre et des Îles Loyauté. Atlas de la Nouvelle-Calédonie. J. Bonvallot. Marseille
(FRA), Nouméa, IRD, Congrès de la Nouvelle-Calédonie: 33-36.
Polidoro, B. A., K. E. Carpenter, L. Collins, N. C. Duke, A. M. Ellison, J. C. Ellison, E. J.
Farnsworth, E. S. Fernando, K. Kathiresan, N. E. Koedam, S. R. Livingstone, T. Miyagi, G.
E. Moore, N. N. Vien, J. E. Ong, J. H. Primavera, S. G. Salmo, J. C. Sanciangco, S. Sukardjo,
Y. M. Wang, J. W. H. Yong (2010). "The loss of species: mangrove extinction risk and
geographic areas of global concern." Plos One 5(4).
Pothin, K., C. Gonzalez-Salas, P. Chabanet, R. Lecomte-Finiger (2006). "Distinction between
Mulloidichthys flavolineatus juveniles from Reunion Island and Mauritius Island (south-west
Indian Ocean) based on otolith morphometrics." Journal of Fish Biology 69(1): 38-53.
R Development Core Team (2011). R: A language and environment for statistical computing.
Vienna, Austria, R Foundation for Statistical Computing.
Ranaldi, M. M., M. M. Gagnon (2008). "Zinc incorporation in the otoliths of juvenile pink
snapper (Pagrus auratus Forster): The influence of dietary versus waterborne sources."
Journal of Experimental Marine Biology and Ecology 360(1): 56-62.
Reis-Santos, P., S. E. Tanner, T. S. Elsdon, H. N. Cabral, B. M. Gillanders (2013). "Effects of
temperature, salinity and water composition on otolith elemental incorporation of
187
Dicentrarchus labrax." Journal of Experimental Marine Biology and Ecology 446(0): 245252.
Rooker, J. R., D. H. Secor, V. S. Zdanowicz, G. De Metrio, L. O. Relini (2003).
"Identification of Atlantic bluefin tuna (Thunnus thynnus) stocks from putative nurseries
using otolith chemistry." Fisheries Oceanography 12(2): 75-84.
Rooker, J. R., G. W. Stunz, S. A. Holt, T. J. Minello (2010). "Population connectivity of red
drum in the northern Gulf of Mexico." Marine Ecology Progress Series 407: 187-196.
Salas, E., H. Molina-Urena, R. P. Walter, D. D. Heath (2010). "Local and regional genetic
connectivity in a Caribbean coral reef fish." Marine Biology 157(2): 437-445.
Sale, P. F. (1971). "Extremely limited home range in a coral reef fish, Dascyllus aruanus
(Pisces; Pomacentridae)." Copeia 1971(2): 324-327.
Sale, P. F. (1991). The ecology of fishes on coral reefs. New York, Academic Press.
Sale, P. F. (2002). Coral Reef Fishes: Dynamics and Diversity in a Complex Ecosystem. San
Diego, Academic Press.
Sale, P. F., R. K. Cowen, B. S. Danilowicz, G. P. Jones, J. P. Kritzer, K. C. Lindeman, S.
Planes, N. V. C. Polunin, G. R. Russ, Y. J. Sadovy, R. S. Steneck (2005). "Critical science
gaps impede use of no-take fishery reserves." Trends in Ecology & Evolution 20(2): 7480.
Sale, P. F., H. Van Lavieren, M. C. Ablan Lagman, J. Atema, M. Butler, C. Fauvelot, J. D.
Hogan, G. P. Jones, K. C. Lindeman, C. B. Paris, R. Steneck, H. L. Stewart (2010).
Preserving Reef Connectivity: A Handbook for Marine Protected Area Managers,
Connectivity Working Group, Coral Reef Targeted Research & Capacity Building for
Management Program, UNU-INWEH.
Sandin, S. A., J. E. Smith, E. E. DeMartini, E. A. Dinsdale, S. D. Donner, A. M. Friedlander,
T. Konotchick, M. Malay, J. E. Maragos, D. Obura, O. Pantos, G. Paulay, M. Richie, F.
Rohwer, R. E. Schroeder, S. Walsh, J. B. C. Jackson, N. Knowlton, E. Sala (2008). "Baselines
and Degradation of Coral Reefs in the Northern Line Islands." Plos One 3(2): 11.
Searcy, S. P., S. Sponaugle (2000). "Variable larval growth in a coral reef fish." Marine
Secor, D. H. (1992). "Application of otolith microchemistry analysis to investigate anadromy
in Chesapeake bay striped bass Morone saxatilis." Fishery Bulletin 90(4): 798-806.
Secor, D. H., J. R. Rooker (2000). "Is otolith strontium a useful scalar of life cycles in
estuarine fishes?" Fisheries Research 46(1-3): 359-371.
188
Secor, D. H., J. R. Rooker, E. Zlokovitz, V. S. Zdanowicz (2001). "Identification of riverine,
estuarine, and coastal contingents of Hudson River striped bass based upon otolith elemental
fingerprints." Marine Ecology-Progress Series 211: 245-253.
Sheaves, M. (2005). "Nature and consequences of biological connectivity in mangrove
systems." Marine Ecology Progress Series 302: 293-305.
Sheaves, M., R. Baker, R. Johnston (2006). "Marine nurseries and effective juvenile habitats:
an alternative view." Marine Ecology-Progress Series 318: 303-306.
Shima, J. S., S. E. Swearer (2009). "Larval quality is shaped by matrix effects: implications
for connectivity in a marine metapopulation." Ecology 90(5): 1255-1267.
Siegel, S., N. J. Castellan (1988). Non parametric statistics for the behavioral sciences. New
York, McGraw-Hill International.
Sigura, A. (2009). La microchimie des otolithes : un outil pour étudier la connectivité entre
écosystèmes du lagon calédonien, Université Bordeaux I: 25.
Soto, C. G., Burhanuddin (1995). "Clove oil as a fish anaesthetic for measuring length and
weight of rabbitfish (Siganus lineatus)." Aquaculture 136(1-2): 149-152.
Spalding, M. D., F. Blasco, C. D. Field (1997). World Mangrove Atlas. Okinawa, Japan.
Standish, J. D., M. Sheehy, R. R. Warner (2008). "Use of otolith natal elemental signatures as
natural tags to evaluate connectivity among open-coast fish populations." Marine Ecology
Progress Series 356: 259-268.
Sturrock, A. M., C. N. Trueman, A. M. Darnaude, E. Hunter (2012). "Can otolith elemental
chemistry retrospectively track migrations in fully marine fishes?" Journal of Fish Biology
81(2): 766-795.
Swearer, S. E., G. E. Forrester, M. A. Steele, A. J. Brooks, D. W. Lea (2003). "Spatiotemporal and interspecific variation in otolith trace-elemental fingerprints in a temperate
estuarine fish assemblage." Estuarine Coastal and Shelf Science 56(5-6): 1111-1123.
Taylor, P. D., L. Fahrig, K. Henein, G. Merriam (1993). "Connectivity is a vital element of
landscape structure." Oikos 68(3): 571-573.
Testau, J. L., F. Conand (1983). Estimation des surfaces des différentes zones des lagons de
Nouvelle-Calédonie. Nouméa, ORSTOM: 10.
Thollot, P. (1992a). "Importance des mangroves pour la faune ichtyologique des récifs
coralliens de Nouvelle-Calédonie." Cybium 16(4): 331-344.
189
Thollot, P. (1992b). Les poissons de mangrove du lagon sud-ouest de Nouvelle-Calédonie.
Centre d'Océanologie de Marseille. Marseille, Université d'Aix-Marseille II. Doctorat: 406.
Thollot, P., M. Kulbicki (1988). Overlap between the fish fauna inventories of coral reefs, soft
bottoms and mangroves in Saint-Vincent Bay (New Caledonia). 6th International Coral Reef
Symposium, Australia.
Thorrold, S. R., C. M. Jones, S. E. Campana (1997). "Response of otolith microchemistry to
environmental variations experienced by larval and juvenile Atlantic croaker (Micropogonias
undulatus)." Limnology and Oceanography 42(1): 102-111.
Thorrold, S. R., G. P. Jones, M. E. Hellberg, R. S. Burton, S. E. Swearer, J. E. Neigel, S. G.
Morgan, R. R. Warner (2002). "Quantifying larval retention and connectivity in marine
populations with artificial and natural markers." Bulletin of Marine Science 70(1): 291-308.
Thorrold, S. R., G. P. Jones, S. Planes, J. A. Hare (2006). "Transgenerational marking of
embryonic otoliths in marine fishes using barium stable isotopes." Canadian Journal of
Thorrold, S. R., D. C. Zacherl, L. A. Levin (2007). "Population connectivity and larval
dispersal using geochemical signatures in calcified structures." Oceanography 20(3): 80-89.
Thresher, R. E., E. B. Brothers (1989). "Evidence of intra-oceanic and inter-oceanic regional
differences in the early life-history of reef-associated fishes." Marine Ecology Progress Series
57(2): 187-205.
Tischendorf, L., L. Fahring (2000). "On the usage and measurement of landscape
connectivity." Oikos 90(1): 7-19.
Totland, M., I. Jarvis, K. E. Jarvis (1992). "An assessment of dissolution techniques for the
analysis of geological samples by plasma spectrometry." Chemical Geology 95: 35-62.
Tournois, J., F. Ferraton, L. Velez, D. J. McKenzie, C. Aliaume, L. Mercier, A. M. Darnaude
(2013). "Temporal stability of otolith elemental fingerprints discriminates among lagoon
nursery habitats." Estuarine, Coastal and Shelf Science 131(0): 182-193.
Tukey, J. W. (1977). Exploratory Data Analysis, Addison-Wesley Publishing Company.
Unsworth, R. K. F., P. S. De Leon, S. L. Garrard, J. Jompa, D. J. Smith, J. J. Bell (2008).
"High connectivity of Indo-Pacific seagrass fish assemblages with mangrove and coral reef
habitats." Marine Ecology Progress Series 353: 213-224.
Unsworth, R. K. F., S. L. Garrard, P. S. De Leon, L. C. Cullen, D. J. Smith, K. A. Sloman, J.
J. Bell (2009). "Structuring of Indo-Pacific fish assemblages along the mangrove-seagrass
continuum." Aquatic Biology 5(1): 85-95.
190
Valiela, I., J. L. Bowen, J. K. York (2001). "Mangrove forests: One of the world's threatened
major tropical environments." BioScience 51(10): 807-815.
Victor, B. C., G. M. Wellington (2000). "Endemism and the pelagic larval duration of reef
fishes in the eastern Pacific Ocean." Marine Ecology Progress Series 205: 241-248.
Vigliola, L., M. G. Meekan (2009). The Back-Calculation of Fish Growth From Otoliths.
Tropical Fish Otoliths: Information for Assessment, Management and Ecology. B. S. Green,
B. D. Mapstone, G. Carlos and G. A. Begg, Springer Netherlands. 11: 174-211.
Vigliola, L., C. Paillon, L. Wantiez (2013). Opération ZoNéCo "Microchimie des otolithes".
ZoNéCo. Nouméa, New Caledonia, IRD-UNC-ZONECO: 74.
Virly, S. (2007). Typologies et biodiversité des mangroves de Nouvelle-Calédonie,
Cartographie des mangroves. ZoNéCo. S. V. Consultant. Nouméa, Nouvelle Calédonie.
Walther, B. D., M. J. Kingsford, M. D. O'Callaghan, M. T. McCulloch (2010). "Interactive
effects of ontogeny, food ration and temperature on elemental incorporation in otoliths of a
coral reef fish." Environmental Biology of Fishes 89(3-4): 441-451.
Walther, B. D., S. R. Thorrold (2006). "Water, not food, contributes the majority of strontium
and barium deposited in the otoliths of a marine fish." Marine Ecology Progress Series 311:
125-130.
Wantiez, L. (2008). "Coral reefs of New Caledonia in 2006: status report and monitoring
network." Revue d'Ecologie - La Terre et la Vie 63(1-2): 117-132.
Wantiez, L., O. Chateau, S. Le Mouellic (2006). "Initial and mid-term impacts of cyclone
Erica on coral reef fish communities and habitat in the South Lagoon Marine Park of New
Caledonia." Journal of the Marine Biological Association of the United Kingdom 86(5):
1229-1236.
Warner, R. R., S. E. Swearer, J. E. Caselle, M. Sheehy, G. Paradis (2005). "Natal traceelemental signatures in the otoliths of an open-coast fish." Limnology and Oceanography
50(5): 1529-1542.
Woodcock, S. H., A. R. Munro, D. A. Crook, B. M. Gillanders (2012). "Incorporation of
magnesium into fish otoliths: Determining contribution from water and diet." Geochimica Et
Cosmochimica Acta 94: 12-21.
Yamashita, Y., T. Otake, H. Yamada (2000). "Relative contributions from exposed inshore
and estuarine nursery grounds to the recruitment of stone flounder, Platichthys bicoloratus,
estimated using otolith Sr : Ca ratios." Fisheries Oceanography 9(4): 316-327.
191
192
Liste des figures
Liste des figures
Figure 1.1 : Représentation schématique des trois composantes de la connectivité (d’après Calabrese & Fagan, 2004). ......................................................................................................... 3
Figure 1.2 : Cycle biologique simplifié de Lutjanus fulviflamma, espèce à œufs pélagiques. . 7
Figure 1.3 : Echelles temporelles et spatiales correspondant aux différentes approches de
mesures de la dispersion des organismes récifaux-coralliens (modifiée de Jones et al. (2009)).
.................................................................................................................................................... 9
Figure 1.4 : Positionnement des otolithes au sein de l’oreille interne d’un téléostéen typique.
(A) Vue dorsale de l’appareil vestibulaire ;; (B) Position des otolithes à l’intérieur du système du labyrinthe (issue de Panfili et al. (2002)). ........................................................................... 12
Figure 1.5 : Morphologie d’une sagitta typique, face interne et externe (A) et détails des trois
plans d’orientations d’une sagitta typique (B) (issue de Panfili et al. (2002)). ....................... 13
Figure 1.6 : Voies d’incorporation des éléments, barrières physiologiques entre l’eau et les otolithes et estimations des taux de transfert pour certains éléments à chaque barrière
physiologique (issue de Campana (1999)). .............................................................................. 15
Figure 2.1 : Situation géographique de la Nouvelle-Calédonie dans le Pacifique Sud-Ouest.21
Figure 2.2 : (A) Sites d’étude : les cercles jaunes et rouges désignent respectivement les sites
à 2 et 4 habitats échantillonnés. (B et C) Détails des habitats échantillonnés au sein des sites à
2 et 4 habitats échantillonnés. (Base de la carte : Millennium Coral Reef Mapping
(Andréfouët & Torres-Pulliza 2004) ; réalisation : M. A. Hamel). .......................................... 23
Figure 2.3 : Détails des quatre sites ateliers: (A) Saint Vincent, (B) Ouano, (C) Gatope et (D)
Prony. ....................................................................................................................................... 25
Figure 2. 4 : Photo des espèces modèles : (A) Lutjanus fulviflamma, (B) Dascyllus aruanus,
(C) Ctenochaetus striatus et (D) Siganus lineatus (Crédits photos Paillon, C. et Randall, J.E.).
.................................................................................................................................................. 29
Figure 2.5 : Illustration des méthodes de prélèvements ; (A) au niveau des mangroves à
l’aide de filets maillants et (B) au niveau des récifs au fusil harpon et à l’eugénol. ................ 30
Figure 2.6 : Protocole de préparation des otolithes pour analyses in toto et analyses en
transects. ................................................................................................................................... 33
Figure 2.7 : Photo d’un système LA-ICP-MS, Environmental Analytical Chemistry Unit,
Research Institute for the Environment and Livelihoods, Charles Darwin University. ........... 35
Figure 2.8 : Illustration de la méthode d’analyse in toto. (A) montage des otolithes entiers sur
lame et (B et C) position de l’analyse laser sur otolithe in toto. .............................................. 37
Figure 2.9 : Illustration de la méthode d’analyse en transects. (A) Coupes transversales
positionnées dans la chambre du LA-ICP-MS et (B) transect sur une coupe transversale
d’otolithe colorée au bleu de Toluidine pour ageage. .............................................................. 38
Figure 2.10 : Protocole de mise en solution des sédiments par acidification. ........................ 39
sites: circles and triangles indicate sites with 2 and 4 habitats sampled, respectively detailed in
C and D (with M: mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier). .... 54
193
Liste des figures
Figure 3.2: Plots of principal component analysis (PCA) of elemental compositions of multispecific otoliths for habitats (A) and stations (B) and plot of contributions of elements in
intermediate reef; BR: inner barrier reef). ................................................................................ 60
Figure 3.3: Plots of principal component analysis (PCA) of elemental compositions of L.
fulviflamma otoliths for habitats (A) and stations (B) and plot of contributions of elements in
intermediate reef; BR: inner barrier reef). ................................................................................ 61
Figure 3.4: Plots of principal component analysis (PCA) of elemental compositions of
otoliths using elements from best combination only (Tab. 3.5) for habitats or stations (A) and
plot of contributions of elements in elemental compositions (B) (M: mangrove; FR: fringing
reef; IR: intermediate reef; BR: inner barrier reef; gat: Gatope, oua: Ouano; pro: Prony; stv: St
Vincent). ................................................................................................................................... 64
Figure 3.5: Plot representing correlations between percent of correct classifications obtained
with mono-specific classifiers and percent of correct RF classifications obtained with
corresponding global (blue circles), functional (red triangles) and taxonomic (green cross)
multi-specific classifiers. .......................................................................................................... 65
Figure 4.1: (A) Location of New Caledonia in the South West Pacific; (B) study sites: circles
and triangles indicate sites sampled with 2 and 4 habitats respectively; (C and D) details of
sites S4 and S2 (M: mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier
reef). ......................................................................................................................................... 81
Figure 4.2: Plots of principal component analysis (PCA) of seawater elemental compositions
barrier reef). .............................................................................................................................. 87
Figure 4.3: Plots of principal component analysis (PCA) of sediment elemental compositions
barrier reef). .............................................................................................................................. 88
sites in New Caledonia: circles indicate sites of otolith microchemistry study ..................... 113
Figure 5.2: Position of the Underwater Visual Census transects (UVC) in the archipelago of
New Caledonia. Polygons indicate the different lagoons, remote reefs and islands for which
we have data; numbers indicate the number of transects. ...................................................... 118
Figure 5.3: Otolith mean value of Ba:Ca, Cr:Ca, Mn:Ca, Rb:Ca, Sn:Ca and Sr:Ca ratios per
habitat, (R) reefs and (M) mangroves. Standard errors (± SE) are plotted. Results of one way
PERMANOVAs are indicated in graphs with (***) p<0.001 and (**) p<0.01. .................... 120
Figure 5.4: PCA biplot of individual fish collected in (M) mangroves and (R) reefs. Only
concentrations of Rb, Sn and Mn in otoliths were used in the PCA. ..................................... 121
Figure 5.5: Prediction of habitats along the juvenile part of otolith transects of adults L.
fulvflamma collected in barrier reefs. Dark grey area represents fish in mangrove whereas
light grey area represents fish with reefs signatures (with M: mangrove and R: reef). ......... 122
Figure 5.6: prediction of juvenile habitat from an otolith transversal section of an adult from
Prony (site 11). ....................................................................................................................... 122
194
Liste des figures
Figure 5.7: Relationship between standardized mangrove area and density of L. fulviflamma
in the different lagoons, reefs and islands of the archipelago of New Caledonia (see map on
Fig. 2) ..................................................................................................................................... 123
sites: Gatope (1), Ouano (2), St Vincent (3) and Prony (4). .................................................. 140
Figure 6.2: Details of habitats sampled in St Vincent (A) Ouano (B), Gatope (C) and Prony
(D). ......................................................................................................................................... 142
Figure 6.3: Otolith mean value of Mn:Ca, Rb:Ca, Sn:Ca Sr:Ca and U:Ca ratios per habitat
(mangroves and reefs). Mean and standard error are plotted. Statistical significances of oneway anova results are indicated above graphs (“***” p-value < 0.001;; “**” p-value < 0.01;
“*” p-value < 0.05;; “+” p-value < 0.1). .................................................................................. 147
Figure 6.4: PCA biplot of individual fish collected in (M) mangroves and (R) all reefs. Only
concentrations of Rb, Pb and Mn in otoliths were used in the PCA. ..................................... 148
Figure 6.5: Representation of the three main patterns of lifespan movement for L.
fulviflamma using otoliths microchemistry and RF predictions. The x axis represents the
otolith radius and the y axis the three possible habitats occupied during life. The second top x
axis indicates the age and length of the fish (respectively below and above axis). The grey
gradient shows the probabilities of habitat predictions (% of vote). The white line joins
habitats predicted with the highest probability at each LA-ICPMS analysis. ........................ 151
195
196
Liste des tableaux
Liste des tableaux
Tableau 2.1: Les onze sites d’échantillonnages (* Sites ateliers;; M : mangrove ; RF : récif
frangeant ; RI : récif intermédiaire ; RB : récif barrière interne). ............................................ 22
Tableau 2.2 : Campagnes d’échantillonnages réalisées et objectifs scientifiques reliés. ....... 26
Table 3.1: Sample size and otolith analyses; (PSO: Pôle de Spectrométrie Océan; RIEL:
Resources Institute of Environmental Livelihoods). ................................................................ 55
Table 3.2: Summary of species characteristic used to build mono-specific signatures (C:
carnivore, MC: micro carnivore, H: herbivore, Z: zooplankton feeders)................................. 55
elemental composition within site and habitat for multi-specific and Lutjanus fulviflamma
(*** p < 0.001; ** p < 0.01; * p < 0.05). ................................................................................. 60
Table 3.4: Summary of RF classification accuracies in percent of individuals correctly
classified to the location they were collected. The highest classification accuracies (≥ 80%) are represented in bold and italic (Gat: Gatope, Oua: Ouano, Pro: Prony, StV: StVincent; M:
mangrove, FR: fringing reef, IR: intermediate reef, BR: barrier reef). .................................... 62
elemental composition using elements from best combination only (Appendix 3) for multispecific at global scale, S. lineatus at regional scale and C. lunulatus at the local scale of
Ouano. ...................................................................................................................................... 63
Table 4.1: Sample size and analyses realized (with M: mangrove; FR: fringing reef; IR:
intermediate reef; BR: inner barrier reef; B: Brest; D: Darwin; M: Montpellier). ................... 82
Table 4.2: Mean temperature (°C) and salinity (S) by habitat with SE: standard error, CV:
coefficient of variation, N: number of station, n: number of measures (M: mangrove, FR:
fringing reef, IR: intermediate reef, BR: barrier reef). ............................................................. 86
between temperature, salinity and chemistries of seawater and sediment (“***” p < 0.001;; “**” p < 0.01;; “*” p < 0.05;; “+” p < 0.1). ................................................................................ 89
between temperature, salinity, environmental chemistry and multi-specific otoliths (“***” p < 0.001;; “**” p < 0.01;; “*” p < 0.05;; “+” p < 0.1)...................................................................... 90
Table 4.5: Summary of RF classification accuracies in percent of environmental samples and
otoliths correctly classified to the location they were collected. The highest classification
accuracies (≥ 80%) are represented in bold and italic (Gat: Gatope, Oua: Ouano, Pro: Prony, StV: StVincent; M: mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier
reef). ......................................................................................................................................... 91
Table 4.6: Comparisons between elements constituting otoliths and environment best
combinations for the highest otoliths RF classification accuracies (correct classification >
80%). Elements contributing to both otolith and environmental signatures are in bold. (M:
mangrove; FR: fringing reef; IR: intermediate reef; BR: inner barrier reef). .......................... 93
Table 5.1: Sampling design in order to: A. characterize reefs and mangrove chemical
signatures by LA-ICPMS of the surface of otoliths and B. determine the juvenile habitat by
LA-ICPMS of transverse sections of otolith along transects from core to 1260 µm ............. 114
197
Liste des tableaux
Table 5.2: RF classification matrix of habitats...................................................................... 120
Table 6.1: Sample size to (A) characterize habitats chemical signatures by LA-ICPMS of the
surface of otoliths and (B) determine the patterns of fish movements by LA-ICPMS of
transverse sections of otolith along transects. ........................................................................ 142
Table 6.2: Random Forest classifications with details of the best combinations of elements
and the percentage of correct classification for each analysis (M: mangrove; FR: fringing reef;
IR: intermediate reef; BR: barrier reef). ................................................................................. 149
198
Annexes
ANNEXES
199
Annexes
200
Annexes
Liste des annexes
Appendix 1: Multivariate results of PERMANOVA examining spatial variation in otolith
elemental composition within site and habitat at the global scale for multi-specific group and
the eight species. .................................................................................................................... 202
Appendix 2: Plots of principal component analysis (PCA) of elemental compositions of
otoliths. ................................................................................................................................... 203
Appendix 3: Recap of the percentage of correct classification obtained by RF and the best
combination of elements for the all classifiers tested. ........................................................... 205
Appendix 4: Percentage of correct classification obtained by RF with mono-specific and
multi-specific classifiers. ........................................................................................................ 206
Appendix 5: Results of Spearman and Mantel tests performed on environmental and otoliths
elemental compositions with salinity ..................................................................................... 207
Appendix 6: Results of Spearman and Mantel tests performed on environmental and otoliths
elemental compositions with temperature .............................................................................. 208
Appendix 7: Results of Spearman and Mantel tests performed on sediment and otoliths
elemental compositions with seawater elemental composition .............................................. 209
Appendix 8: Results of Spearman and Mantel tests performed on seawater and otoliths
elemental compositions with sediment elemental composition ............................................. 210
Appendix 9: Summary of the percentage of correct classification obtained by RF and the best
combination of elements for the all classifiers tested. ........................................................... 211
Appendix 10: Spatial variation in otolith fingerprints of Dascyllus aruanus ....................... 212
Appendix 11: Spatial variation in otolith fingerprints of Lutjanus fulviflamma ................... 213
Appendix 12: Lifetime patterns of movements for Lutjanus fulviflamma ............................ 214
201
Annexes
Appendix 1: Multivariate results of PERMANOVA examining spatial variation in otolith
elemental composition within site and habitat at the global scale for multi-specific group and
the eight species.
Species
df
MS
F
p
Multi-specific
Site
Habitat
Site x Habitat
10
3
16
0.003
0.076
0.003
1.99
49.29
2.19
0.006**
0.001***
0.001***
L. fulviflamma
Site
Habitat
Site x Habitat
10
3
14
0.014
0.005
0.0008
3.27
11.29
1.92
0.001***
0.001***
0.013*
D. aruanus
Site
Habitat
Site x Habitat
9
2
6
0.0023
0.0020
0.0018
1.96
1.68
1.49
0.04*
0.17
0.11
C. striatus
Site
Habitat
Site x Habitat
10
2
6
0.0005
0.0005
0.0005
0.89
0.87
0.98
0.61
0.46
0.46
C. lunulatus
Site
Habitat
Site x Habitat
10
2
6
0.006
0.0003
0.0007
2.15
0.91
2.39
0.026 *
0.44
0.023 *
S. bilineata
Site
Habitat
Site x Habitat
10
1
2
0.006
0.0022
0.0007
2.0620
7.3486
2.3191
0.012*
0.002**
0.063 +
G. aureolineatus
Site
9
0.0006
1.3777
0.113
G. oyena
Site
7
0.0031
3.8955
0.005**
S. lineatus
Site
8
0.0036
4.8142
0.001***
(*** p < 0.001; ** p < 0.01; * p < 0.05; + p < 0.1)
202
Annexes
Appendix 2: Plots of principal component analysis (PCA) of elemental compositions of
otoliths.
A
B
A
B
C
A
B
C
A
B
C
S.#bilineata#
C. lunulatus
C. striatus
D. aruanus
C
Plots of PCA of D. aruanus, C. striatus, C. lunulatus and S. bilineata elemental compositions
of otoliths for habitats (A) and stations (B) and plot of contributions of elements in habitat and
station elemental compositions (C) (with FR: fringing reef; IR: intermediate reef; BR: inner
barrier reef).
203
Annexes
Plots of PCA of of G. aureolineatus, G. oyena and S. lineatus elemental compositions of
otoliths for stations (A) and plot of contributions of elements in habitat and station elemental
compositions (B).
204
Annexes
Appendix 3: Recap of the percentage of correct classification obtained by RF and the best
combination of elements for the all classifiers tested.
Species
Multi-specific
Scale
Local
Regional
Global
L. fulviflamma
Local
Regional
Global
D. aruanus
Local
Regional
C. striatus
Global
Local
Regional
C. lunulatus
Global
Local
Regional
S. bilineata
S. lineatus
G. aureolineatus
G. oyena
Global
Regional
Global
Regional
Global
Regional
Global
Regional
Global
Site
Gatope
Ouano
Prony
St Vincent
Pilot Sites
Pilot Sites
Pilot Sites
Pilot Sites
Pilot Sites
All
All
All
Gatope
Ouano
Prony
St Vincent
Pilot Sites
Pilot Sites
Pilot Sites
Pilot Sites
Pilot Sites
All
All
All
Gatope
Ouano
Prony
St Vincent
Pilot Sites
Pilot Sites
Pilot Sites
Pilot Sites
All
Gatope
Ouano
Prony
St Vincent
Pilot Sites
Pilot Sites
Pilot Sites
Pilot Sites
All
Gatope
Ouano
Prony
St Vincent
Pilot Sites
Pilot Sites
Pilot Sites
Pilot Sites
All
Pilot Sites
Pilot Sites
Pilot Sites
All
Pilot Sites
All
Pilot Sites
All
Pilot Sites
All
Habitat
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M
FR
IR
BR
M/BR
M
BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M
FR
IR
BR
M/BR
M
BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR
IR
BR
BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR
IR
BR
BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR
IR
BR
BR
FR/BR
FR
BR
BR
M
M
BR
BR
M
M
Level
Habitat
Habitat
Habitat
Habitat
Habitat
Stations
Stations
Stations
Stations
Habitat
Stations
Stations
Habitat
Habitat
Habitat
Habitat
Habitat
Stations
Stations
Stations
Stations
Habitat
Stations
Stations
Habitat
Habitat
Habitat
Habitat
Habitat
Stations
Stations
Stations
Stations
Habitat
Habitat
Habitat
Habitat
Habitat
Stations
Stations
Stations
Stations
Habitat
Habitat
Habitat
Habitat
Habitat
Stations
Stations
Stations
Stations
Habitat
Stations
Stations
Stations
Stations
Stations
Stations
Stations
Stations
Stations
Correct classification (%)
63
49
57
69
59
63
42
44
47
90
44
29
74
43
78
73
61
100
63
54
48
95
65
29
74
74
67
85
64
64
67
60
41
79
83
96
71
71
71
64
63
47
67
80
67
80
67
50
58
83
41
95
78
100
37
84
63
80
24
83
67
Combination of elements
Ba Mg Mn Pb Rb Sn Zn
B Ba Mn Pb Sn Th Zn
B Ba Mn Sn U Zn
Ba Mg Mn Pb Rb U
B Ba Cr Mg Mn Rb Sn
B Ba Sn U
B Ba Mg Pb Rb Sn Zn
Ba Rb Sn Sr Zn
B Pb Sn Th U Zn
Ba Mg Mn Pb Rb Sn U
B Ba Cr Mn Sn
B Pb Rb Sn Sr
Ba Cr Mg Mn U Zn
B Mn Pb Zn
B Ba Mg Mn Pb Sn
B Cr Mn Pb Rb U
Cr Mg Mn Pb Rb Sr
Ba Cr Mg Pb Rb Sn Zn
B Mg Mn Pb Rb Sn
Ba Mg Mn Sr U
Cr Mn Pb Sn U Zn
B Cr Mn Pb Rb Sn Th U
B Ba Mn Pb Sn Zn
Ba Cr Sn U Zn
B Ba Cr Mn Rb Sr
B Ba Cr Mg Sn Th
B Ba Mg Mn Rb Sr
B Sn Sr Zn
B Ba Cr Mg Mn Pb Rb Zn
B Mg Sr U
B Ba Th Zn
B Ba Pb Sn U
B Ba Cr Mg Mn Rb Sn Sr Th U
B Mn
B Rb Sn U
B Cr Pb Rb Sn Sr Th Zn
Cr Mg Pb Th
B Cr Mg Mn Pb
Cr Mn Rb Sr Th U
B Ba Rb U
B Mn Sr U
Ba Mg Mn Pb Sn Sr Th
Cr Mn Pb Rb
Ba Mg Mn Rb Zn
B Sn
Ba Mn Rb
Ba Rb Sn
Rb
Rb U
Ba Cr Mg Mn Pb Sr
B Mn Pb Rb Sr
B Mn Pb Sr Th
Mg Pb Zn
Mg Mn
Mn U
B Mn
B Mn Rb U
Ba
Ba Cr Sn Th
B Sn Zn
B Ba Co Cr Mn Ni U
205
Annexes
Appendix 4: Percentage of correct classification obtained by RF with mono-specific and
multi-specific classifiers.
Scale
Local Gatope
Local Gatope
Local Gatope
Local Gatope
Local Gatope
Local Gatope
Local Ouano
Local Ouano
Local Ouano
Local Ouano
Local Ouano
Local Ouano
Local Prony
Local Prony
Local Prony
Local Prony
Local Prony
Local Prony
Local St Vincent
Local St Vincent
Local St Vincent
Local St Vincent
Local St Vincent
Local St Vincent
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Regional
Global
Global
206
Species
C.lunulatus
C.striatus
D.aruanus
D.aruanus
L.fulviflamma
L.fulviflamma
C.lunulatus
C.striatus
D.aruanus
D.aruanus
L.fulviflamma
L.fulviflamma
C.lunulatus
C.striatus
D.aruanus
D.aruanus
L.fulviflamma
L.fulviflamma
C.lunulatus
C.striatus
D.aruanus
D.aruanus
L.fulviflamma
L.fulviflamma
C.lunulatus
C.striatus
D.aruanus
D.aruanus
L.fulviflamma
L.fulviflamma
G.oyena
G.oyena
L.fulviflamma
L.fulviflamma
S.lineatus
C.lunulatus
C.striatus
D.aruanus
D.aruanus
L.fulviflamma
L.fulviflamma
S.bilineata
S.bilineata
C.lunulatus
C.striatus
C.striatus
D.aruanus
D.aruanus
L.fulviflamma
L.fulviflamma
C.lunulatus
C.striatus
C.striatus
D.aruanus
D.aruanus
G.aureolineatus
G.aureolineatus
L.fulviflamma
L.fulviflamma
S.bilineata
S.bilineata
L.fulviflamma
L.fulviflamma
Habitat
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M
M
M
M
M
FR
FR
FR
FR
FR
FR
FR
FR
IR
IR
IR
IR
IR
IR
IR
BR
BR
BR
BR
BR
BR
BR
BR
BR
BR
BR
M/BR
M/BR
Group
global
global
global
taxa
global
diet
global
global
global
taxa
global
diet
global
global
global
taxa
global
diet
global
global
global
taxa
global
diet
global
global
global
taxa
global
diet
global
diet
global
diet
global
global
global
global
taxa
global
diet
global
diet
global
global
diet
global
taxa
global
diet
global
global
diet
global
taxa
global
diet
global
diet
global
diet
global
diet
Mono-specific (%)
67
79
74
74
74
74
80
83
74
74
43
43
67
96
67
67
78
78
80
71
85
85
73
73
67
71
64
64
67
67
83
83
100
100
84
50
71
64
64
63
63
78
78
58
64
64
67
67
54
54
83
63
63
60
60
80
80
48
48
100
100
79
79
Multi-specific (%)
50
32
41
31
29
29
22
48
17
24
26
29
22
46
33
37
22
26
42
42
52
23
49
58
52
29
35
32
40
39
25
0
20
11
44
8
32
31
44
28
32
22
0
17
39
39
38
33
26
23
46
18
39
35
42
10
20
31
35
0
17
79
51
Annexes
Appendix 5: Results of Spearman and Mantel tests performed on environmental and otoliths elemental compositions with salinity
(“***” p < 0.001;; “**” p < 0.01;; “*” p < 0.05;; “+” p < 0.1)
Seawater
Sediment
Multispecific
L. fulviflamma
D. aruanus
C. striatus
C. lunulatus
S. bilineata
G.aureolineatus
S. lineatus
G. oyena
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
B
-0.56
**
0.32
.
-0.40
*
-0.16
NS
0.26
NS
-0.34
NS
0.06
NS
-0.29
NS
-0.63
.
0.44
NS
-0.11
NS
Ba
-0.56
**
-0.11
NS
-0.31
NS
-0.17
NS
-0.38
NS
-0.14
NS
0.24
NS
0.24
NS
0.10
NS
0.06
NS
-0.38
NS
Cr
-0.05
NS
0.02
NS
-0.37
*
-0.51
**
-0.29
NS
0.44
.
0.19
NS
-0.04
NS
0.03
NS
-0.49
NS
-0.76
*
Mg
0.41
*
-0.35
.
-0.38
*
-0.26
NS
0.10
NS
0.04
NS
0.11
NS
-0.18
NS
0.34
NS
0.54
NS
0.05
NS
Mn
-0.23
NS
-0.06
NS
-0.35
.
0.11
NS
0.42
.
0.31
NS
0.04
NS
0.47
NS
0.25
NS
0.56
NS
0.14
NS
Pb
-0.38
*
-0.15
NS
-0.27
NS
-0.29
NS
-0.18
NS
-0.33
NS
0.30
NS
-0.02
NS
-0.12
NS
0.04
NS
-0.11
NS
Rb
0.29
NS
-0.09
NS
0.10
NS
0.11
NS
0.28
NS
-0.09
NS
-0.04
NS
-0.12
NS
-0.25
NS
-0.10
NS
-0.54
NS
Sn
-0.32
NS
-0.05
NS
-0.56
**
-0.35
.
0.08
NS
0.23
NS
-0.19
NS
0.18
NS
-0.25
NS
0.41
NS
-0.18
NS
Sr
0.02
NS
0.34
.
0.42
*
0.31
NS
0.10
NS
0.19
NS
-0.14
NS
0.07
NS
0.08
NS
-0.32
NS
-0.32
NS
Th
-0.04
NS
-0.15
NS
0.07
NS
0.11
NS
-0.01
NS
-0.37
NS
-0.07
NS
-0.02
NS
-0.13
NS
NA
NA
NA
NA
U
-0.54
**
0.10
NS
-0.24
NS
-0.19
NS
-0.33
NS
-0.03
NS
0.06
NS
0.53
.
0.72
*
0.12
NS
0.34
NS
Zn
0.07
NS
-0.09
NS
-0.41
*
-0.51
**
0.12
NS
-0.17
NS
0.13
NS
0.48
.
0.14
NS
0.67
.
-0.41
NS
All
0.10
NS
-0.10
NS
0.11
NS
0.23
.
-0.15
NS
0.30
*
-0.15
NS
-0.08
NS
-0.07
NS
-0.07
NS
-0.27
NS
207
Annexes
Appendix 6: Results of Spearman and Mantel tests performed on environmental and otoliths elemental compositions with temperature
(“***” p < 0.001;; “**” p < 0.01;; “*” p < 0.05;; “+” p < 0.1, NS: non significant, NA: no data)
Seawater
Sediment
Multispecific
L. fulviflamma
D. aruanus
C. striatus
C. lunulatus
S. bilineata
G.aureolineatus
S. lineatus
G. oyena
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
B
0.35
.
-0.06
NS
-0.20
NS
-0.60
**
0.50
*
0.34
NS
-0.18
NS
0.19
NS
-0.26
NS
-0.12
NS
0.32
NS
Ba
0.20
NS
-0.14
NS
0.18
NS
0.10
NS
0.44
.
-0.06
NS
0.14
NS
0.34
NS
0.32
NS
0.70
NS
-0.29
NS
Cr
-0.43
*
-0.14
NS
-0.62
***
-0.36
.
-0.04
NS
-0.52
*
-0.03
NS
-0.31
NS
-0.10
NS
-0.61
NS
-0.32
NS
Mg
-0.38
.
-0.09
NS
-0.49
**
-0.40
*
-0.33
NS
-0.18
NS
-0.48
*
0.17
NS
-0.29
NS
0.06
NS
-0.46
NS
Mn
-0.32
NS
-0.15
NS
-0.34
.
-0.59
**
-0.17
NS
-0.43
.
-0.43
.
-0.07
NS
-0.25
NS
-0.38
NS
-0.17
NS
Pb
0.44
*
-0.29
NS
0.26
NS
-0.08
NS
0.02
NS
0.55
*
0.02
NS
0.64
*
-0.04
NS
0.12
NS
-0.38
NS
Rb
-0.40
*
-0.18
NS
0.29
NS
0.53
**
0.17
NS
0.59
**
0.15
NS
-0.21
NS
-0.31
NS
-0.29
NS
0.46
NS
Sn
0.22
NS
-0.31
NS
0.09
NS
-0.16
NS
-0.46
.
-0.21
NS
0.02
NS
-0.21
NS
-0.27
NS
0.52
NS
-0.41
NS
Sr
0.28
NS
0.08
NS
0.39
*
0.42
*
-0.09
NS
-0.03
NS
0.43
.
-0.32
NS
0.23
NS
-0.23
NS
0.23
NS
Th
-0.44
*
-0.24
NS
0.06
NS
-0.07
NS
0.24
NS
0.06
NS
-0.13
NS
0.06
NS
-0.60
.
NA
NA
NA
NA
U
0.27
NS
-0.09
NS
0.50
**
0.46
*
0.42
.
0.28
NS
-0.22
NS
0.37
NS
-0.61
.
-0.43
NS
-0.19
NS
Zn
0.11
NS
-0.23
NS
-0.46
*
0.22
NS
-0.53
*
0.02
NS
-0.03
NS
0.12
NS
-0.27
NS
-0.06
NS
0.46
NS
All
0.04
NS
-0.10
NS
0.16
.
0.34
**
0.10
NS
0.02
NS
0.12
NS
-0.06
NS
0.44
.
-0.26
NS
-0.35
NS
208
Annexes
Appendix 7: Results of Spearman and Mantel tests performed on sediment and otoliths elemental compositions with seawater elemental composition
Sediment
Multispecific
L. fulviflamma
D. aruanus
C. striatus
C. lunulatus
S. bilineata
G.aureolineatus
S. lineatus
G. oyena
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
B
0.00
NS
0.10
NS
-0.34
.
0.29
NS
0.29
NS
-0.15
NS
0.53
.
0.07
NS
-0.88
**
-0.12
NS
Ba
-0.07
NS
0.23
NS
0.29
NS
0.45
.
0.36
NS
-0.04
NS
-0.12
NS
0.12
NS
0.45
NS
0.74
*
Cr
0.25
NS
0.27
NS
0.20
NS
-0.08
NS
0.22
NS
0.11
NS
-0.32
NS
-0.10
NS
0.51
NS
0.55
NS
Mg
-0.16
NS
-0.19
NS
0.01
NS
0.08
NS
0.24
NS
0.07
NS
0.09
NS
0.18
NS
0.53
NS
0.55
NS
Mn
-0.07
NS
0.85
***
0.62
***
0.51
*
0.00
NS
-0.31
NS
0.46
.
0.24
NS
-0.03
NS
0.57
NS
Pb
-0.04
NS
0.36
.
0.18
NS
-0.13
NS
0.52
*
0.12
NS
0.53
.
-0.07
NS
0.20
NS
0.16
NS
Rb
0.19
NS
-0.43
*
-0.23
NS
-0.07
NS
-0.17
NS
-0.16
NS
0.34
NS
-0.36
NS
-0.43
NS
-0.57
NS
Sn
0.19
NS
0.11
NS
-0.12
NS
-0.06
NS
0.10
NS
-0.07
NS
-0.22
NS
-0.23
NS
0.20
NS
-0.08
NS
Sr
0.31
.
-0.12
NS
0.21
NS
-0.55
*
0.02
NS
-0.06
NS
-0.21
NS
-0.50
NS
0.18
NS
-0.24
NS
Th
0.23
NS
0.05
NS
0.33
.
-0.19
NS
0.08
NS
-0.31
NS
-0.12
NS
0.01
NS
NA
NA
NA
NA
U
-0.20
NS
0.08
NS
-0.09
NS
0.40
NS
0.25
NS
-0.04
NS
-0.32
NS
-0.20
NS
0.70
*
-0.35
NS
Zn
0.06
NS
0.19
NS
0.30
NS
0.01
NS
-0.34
NS
0.32
NS
0.43
NS
0.14
NS
-0.55
NS
-0.52
NS
All
-0.04
NS
0.16
NS
-0.02
NS
-0.02
NS
0.10
NS
-0.04
NS
-0.03
NS
-0.18
NS
0.01
NS
0.50
*
209
Annexes
Appendix 8: Results of Spearman and Mantel tests performed on seawater and otoliths elemental compositions with sediment elemental composition
Seawater
Multispecific
L. fulviflamma
D. aruanus
C. striatus
C. lunulatus
S. bilineata
G.aureolineatus
S. lineatus
G. oyena
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
Rs
p
B
0.00
NS
-0.25
NS
0.02
NS
0.19
NS
-0.07
NS
0.05
NS
-0.09
NS
-0.02
NS
-0.03
NS
-0.50
NS
Ba
-0.07
NS
-0.20
NS
-0.13
NS
0.08
NS
0.39
NS
-0.49
*
-0.31
NS
-0.08
NS
-0.37
NS
-0.64
.
Cr
0.25
NS
0.05
NS
0.06
NS
0.07
NS
0.10
NS
-0.23
NS
0.07
NS
-0.44
NS
-0.15
NS
0.29
NS
Mg
-0.16
NS
-0.16
NS
-0.13
NS
-0.38
NS
0.24
NS
-0.11
NS
0.06
NS
-0.08
NS
-0.08
NS
0.10
NS
Mn
-0.07
NS
-0.08
NS
0.00
NS
0.08
NS
0.41
.
-0.06
NS
0.09
NS
-0.13
NS
0.10
NS
-0.57
NS
Pb
-0.04
NS
-0.07
NS
-0.27
NS
-0.08
NS
0.28
NS
0.15
NS
-0.51
.
0.68
*
0.05
NS
-0.63
.
Rb
0.19
NS
-0.27
NS
0.05
NS
0.34
NS
-0.20
NS
-0.53
*
0.39
NS
0.43
NS
0.30
NS
-0.67
.
Sn
0.19
NS
0.12
NS
0.34
.
-0.04
NS
-0.27
NS
0.00
NS
0.03
NS
0.39
NS
0.10
NS
-0.62
NS
Sr
0.31
.
-0.10
NS
-0.01
NS
-0.34
NS
0.45
.
-0.16
NS
-0.05
NS
-0.09
NS
0.00
NS
-0.26
NS
Th
0.23
NS
0.09
NS
0.24
NS
0.38
NS
0.26
NS
-0.15
NS
-0.13
NS
0.60
.
NA
NA
NA
NA
U
-0.20
NS
-0.06
NS
-0.21
NS
-0.03
NS
0.28
NS
0.47
*
0.34
NS
-0.10
NS
-0.07
NS
0.00
NS
Zn
0.06
NS
-0.01
NS
-0.01
NS
0.13
NS
-0.07
NS
-0.12
NS
-0.25
NS
0.20
NS
-0.57
NS
-0.60
NS
All
-0.04
NS
0.46
*
0.52
*
-0.03
NS
0.05
NS
-0.11
NS
-0.07
NS
0.00
NS
-0.20
NS
0.51
*
210
Annexes
Appendix 9: Summary of the percentage of correct classification obtained by RF and the best
combination of elements for the all classifiers tested.
Species
Scale
Site
Habitat
Level
Seawater
Local
Gatope
Ouano
Prony
St Vincent
West coast
West coast
West coast
West coast
West coast
All
All
All
Gatope
Ouano
Prony
St Vincent
West coast
West coast
West coast
West coast
West coast
All
All
All
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M
FR
IR
BR
M/BR
M
BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M/FR/IR/BR
M
FR
IR
BR
M/BR
M
BR
Habitat
Habitat
Habitat
Habitat
Habitat
Station
Station
Station
Station
Habitat
Station
Station
Habitat
Habitat
Habitat
Habitat
Habitat
Station
Station
Station
Station
Habitat
Station
Station
Regional
Global
Sediment
Local
Regional
Global
Correct
classification (%)
79
67
50
71
74
86
33
50
35
95
63
22
63
88
50
43
42
25
50
75
57
76
23
30
Combination of elements
Ba Cr Sn Zn
B Mg Mn
Ba Mn U
Mn Pb
Mg Mn Rb Sn Sr
Cr Mg Th
Th
Zn
Ba Pb
Mg Mn Zn
Cr Mg Mn Zn
Pb
Zn
Cr
Pb U
B
Mg Mn Pb Sn
Cr
Pb
Pb
B Cr Sr
B Mg Sr
Pb
Cr Sr
(M: mangrove, FR: fringing reef, IR: intermediate reef, BR: barrier reef).
211
Annexes
Appendix 10: Spatial variation in otolith fingerprints of Dascyllus aruanus
Multivariate PERMANOVA showed a statistical significant difference in otolith
chemistry for fish from Ouano and St Vincent reefs (p-value = 0.04 and 0.02 respectively).
There was no significant difference for otoliths from Gatope and Prony reefs (p-value = 0.6
and 0.8 respectively).
One-way ANOVA revealed significant differences in concentrations of B, Ba, Mg, Rb, Sr
and Zn and post-hoc tests revealed between fringing, intermediate and barrier reefs which
ones displayed these differences. These differences resulted in the following reef groups:
intermediate and barrier reefs were pooled in the offshore reefs and fringing reefs constituted
the inshore reefs in Gatope, Ouano and Prony. On the contrary, fringing and intermediate
reefs were pooled in the inshore reef and barrier reef constituted the offshore reef in St
Vincent.
Dascyllus aruanus otolith elemental concentrations of B:Ca, Ba:Ca, Mg:Ca, Rb:Ca, Sr:Ca and
Zn:Ca per reef habitat. Mean and standard error are plotted. Bar labelled with different letters
(a, b, c) are statistically different from each over (Non parametric post-hoc tests, multiple
comparisons of mean ranks for all groups (Siegel and Castellan, 1988)).
212
Annexes
Appendix 11: Spatial variation in otolith fingerprints of Lutjanus fulviflamma
Multivariate PERMANOVA showed a statistical significant difference between
elemental compositions of otoliths inshore and offshore reefs from St Vincent (p-value =
0.03). There were no significant differences observed between reefs inshore and offshore
reefs of Gatope, Ouano and Prony (p-value = 0.8, 0.9 and 0.2 respectively).
One-way ANOVA revealed significant differences in B, Ba Mn and Sr concentrations in
otoliths between reefs. In Gatope, Mn concentrations were higher and B concentrations were
lower in inshore reef compared to offshore reef. In Ouano, elemental concentrations of
otoliths showed no significant differences between inshore and offshore reefs. In Prony, B
and Mn concentrations were higher and Ba concentrations were lower in inshore reefs
compared to offshore reefs. In St Vincent, concentrations of Ba and Sr were lower in inshore
reefs compared to offshore reefs.
Lutjanus fulviflamma otolith elemental concentrations of B:Ca, Ba:Ca, Mn:Ca and Sr:Ca per
reef habitat (coastal vs large reefs). Mean and standard error are plotted. Statistical
significances of one-way anova results are indicated above graphs (“***” p-value < 0.001;
“**” p-value < 0.01;; “*” p-value < 0.05).
213
Annexes
Appendix 12: Lifetime patterns of movements for Lutjanus fulviflamma
The first pattern showed four individuals evolving in two different habitats, mangrove as
young then offshore reef as adult.
Representation of the first pattern of lifespan movement for L. fulviflamma using otoliths
microchemistry and RF predictions. The x axis represents the otolith radius and the y axis the
three possible habitats occupied during life. The second top x axis indicates the age and length
of the fish (respectively below and above axis). The grey gradient shows the probabilities of
habitat predictions (% of vote). The white line joins habitats predicted with the highest
probability at each LA-ICPMS analysis
214
Annexes
The second pattern showed five individuals evolving in three different habitats: mangrove as
young, then inshore and offshore reefs as adults with no return to mangrove for adults but
random movements between inshore and offshore reefs (Fig. 5).
Representation of the second pattern of lifespan movement for L. fulviflamma using otoliths
215
Annexes
The third pattern showed eleven individuals evolving in three different habitats and presenting
random movements between them: mangrove as young, then inshore and offshore reefs as
adults with return to mangrove for adults (Fig. 6a and 6b).
Representation of the third pattern of lifespan movement for L. fulviflamma using otoliths
216
Annexes
Representation of the third pattern of lifespan movement for L. fulviflamma using otoliths
probability at each LA-ICPMS analysis.
217
Résumé -
La connectivité écologique se mesure via l’estimation des mouvements réalisés par les organismes au cours de leur vie. Parmi les outils de mesure existants, seule la microchimie des
otolithes est capable de reconstruire les mouvements des poissons au cours de leur vie entière,
notamment lors des migrations ontogénétiques. Son utilisation au sein des milieux oligotrophes et
chimiquement peu contrastés tels que les récifs coralliens reste cependant marginale. C’est dans ce contexte que s’inscrit ce travail sur la microchimie des otolithes des poissons des récifs coralliens et mangroves de Nouvelle-Calédonie. Il se divise en deux parties. La première, méthodologique, est
axée sur l’estimation du potentiel de l’outil microchimie des otolithes en Nouvelle-Calédonie, avec
un chapitre sur la détermination des signatures multi-élémentaires caractéristiques des différents
habitats de la Grande Terre et un chapitre sur l’étude de la relation entre la microchimie de l’environnement et celle des otolithes. La seconde partie se concentre sur l’application de la méthodologie précédemment développée à des thématiques écologiques, avec un chapitre sur le rôle
des mangroves sur le cycle de vie et la distribution géographique d’une espèce de Lutjanidae, Lutjanus fulviflamma, et un chapitre centré sur les mouvements réalisés au cours de la vie de cette
espèce. Les résultats démontrent que la microchimie des otolithes présente un potentiel élevé en
Nouvelle-Calédonie avec un fort pouvoir de discrimination entre la mangrove et les récifs
coralliens. La relation entre la microchimie de l’environnement et celle des otolithes est complexe et dépend fortement des espèces considérées. Cependant, les contrastes chimiques marqués de
l’environnement se retrouvent au sein des otolithes. Les résultats démontrent une connexion forte entre différents habitats pour L. fulviflamma, particulièrement entre la mangrove et les récifs
coralliens avec une importance cruciale de la mangrove comme zone de nurserie. Trois différents
patrons de mouvements entre habitats ont été identifiés. Cette diversité illustre une forte variabilité
de la connectivité écologique entre les individus, avec pour certains d’entre eux, une fréquentation de l’habitat juvénile durant la phase adulte. Mots clés : microchimie des otolithes ; poissons des récifs coralliens ; connectivité ; Lutjanus
fulviflamma ; mangrove ; nurserie ; patrons de mouvements.
Abstract -
Ecological connectivity is defined by organism movements between habitats.
Among the tools used to measure connectivity, otolith microchemistry is the only one able to
reconstruct the fish movement throughout their entire lifetime, and thus include ontogenetic
migrations. However, it is seldom used in oligotrophic environments that typically show poor
chemical contrasts such as coral reefs. In this context, this study focused on otolith microchemistry
of coral reef and mangrove fishes of New Caledonia. This work comprises two parts. First, a
methodological part to assess the potential of otolith microchemistry in New Caledonia, with one
chapter on the determination of multi-elemental signatures of habitats from the Main Island and one
chapter on the relationships between environmental and otoliths microchemistries. Second, the
previously developed methodology was applied to ecological questions, with a chapter on the
importance of mangrove on the life cycle and the spatial distribution of a Lutjanidae, Lutjanus
fulviflamma, and a chapter on the lifetime movement patterns of this species. Results revealed a
high potential of otolith microchemistry in New Caledonia with a high power of discrimination
between mangroves and coral reefs. Relationships between environmental and otolith
microchemistries were complex and species-specific. However, strong chemical contrasts in the
environment were traduced in the otoliths. Results showed a strong connection between different
habitats for L. fulviflamma, in particular between mangroves and reefs with a crucial importance of
mangroves as nursery grounds. Three different patterns of lifetime movements among habitats were
identified. This diversity illustrates a high variability of ecological connectivity patterns among
individuals, with a possible return to the juvenile habitat during the adult stage.
Key words: otolith microchemistry; coral reef fish; connectivity; Lutjanus fulviflamma; mangrove;
nursery; movement patterns.

Etude de la connectivité entre les communautés de poissons de

Transcription

Similar documents

Etude sedimentologique des milieux de mangrove du Benin

Groupe d`Etude des Milieux Estuariens et Littoraux

pavillon tropical

Black Book: The Tragedy of Pontus, 1914-1922

Insectarium de La Réunion

Ultrastructural studies on the merogony of schellackia cf. agamae

anglais : test de positionnement (45 minutes)

Wake up! - Firkowski

Home-school advantage